Lead Exposure, Homocysteine, Dna Methylation and Late-Onset Alzheimer’S Disease

LEAD EXPOSURE, HOMOCYSTEINE, DNA METHYLATION AND LATE-ONSET ALZHEIMER’S DISEASE

Kelly M. Bakulski

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Environmental Health Sciences) in The University of Michigan 2012

Doctoral Committee:

Professor Howard Hu, Co-Chair Assistant Professor Laura S. Rozek, Co-Chair Professor Henry L. Paulson Assistant Professor Sung Kyun Park

ACKNOWLEDGMENTS

This dissertation was written by Kelly M. Bakulski, with extensive contributions by Drs. Howard Hu, Dana Dolinoy, Laura Rozek, Henry Paulson, Sung Kyun Park, Maureen Sartor, Bhramar Mukherjee, and Jennifer Weuve. I would like to thank my dissertation committee members, Howard, Laura, Hank, and Sung Kyun for the extensive time and mentorship they provided me. I would also like to acknowledge Dr. Mukherjee for serving on my preliminary committee and pointing me in the right direction from the start. Dana and Maureen also provided extremely valuable guidance and mentorship. I feel very lucky to have had this wide net of professional support. I had the wonderful opportunity to have worked in the collaborative Hu, Dolinoy, and Rozek laboratories and I am very grateful to my many, many labmates for their assistance and advice. My Alzheimer’s project was supported by the folks at the Michigan Alzheimer’s Research Center, including Hank, Dr. Sid Gilman, Dr. Andrew Lieberman, Dr. Roger Albin, Dr. Bruno Giordani, Arijit Bhaumik, Sherry Teboe, and Lisa Bain. The Normative Aging Study project was mentored by Jennifer, Sung Kyun, and Dr. Kathleen Tucker. I would like to extend special thanks to my family for their support, in particular my mom, Barbara, my dad, David, my sister, Susan, and my grandmother, Irene. I would also like to acknowledge the encouragement from my Pauline roommates through the years, including Justin, Ashley, Murph, Brenna, Heather, Lauren, Jon, and Luis.

ii TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii LIST OF FIGURES v LIST OF TABLES vii LIST OF ABBREVIATIONS ix ABSTRACT x

CHAPTER I. Introduction: Alzheimer’s disease and environmental exposure to 1 lead: The epidemiologic evidence and potential role of epigenetics Abstract 1 Alzheimer’s disease 2 General Alzheimer’s disease epidemiology 2 Lead (Pb) exposure 5 Overview of Pb exposure as a risk factor 5 Pb exposure epidemiology 5 Biomarkers of Pb exposure 8 Pb as a neurotoxicant and risk factor for late-onset AD 8 Overview of epigenetics 10 Epigenetic epidemiology and Alzheimer’s disease 10 Epigenetics and heavy metals, with a focus on Pb 13 Data integrating Alzheimer’s disease, epigenetics and Pb 14 exposure Alzheimer’s disease and Pb exposure are associated with 14 changes in one-carbon metabolism, the substrate for DNA methylation Animal model studies linking Pb exposure, epigenetics, and 16 amyloidogenesis Challenges to LOAD epidemiologic research integrating 17 epigenetics and Pb exposure Potential approaches to study Pb exposure, epigenomics, and 18 Alzheimer’s disease epidemiology Figures 21 References 22 II. Research Chapter 1: Lead exposure, B-vitamins, and plasma 39 homocysteine in the VA Normative Aging Study Abstract 39 Introduction 41 Methods 45 Results 49

iii Discussion 52 Tables 57 Figures 64 References 68 Supplemental Tables 72 Supplemental Figures 76 III. Research Chapter 2: Genome-wide DNA methylation differences 77 between late-onset Alzheimer’s disease and cognitively normal controls in the human frontal cortex. Abstract 77 Introduction 79 Methods 81 Results 85 Discussion 91 Figures 97 Tables 104 References 108 IV. Research Chapter 3: An integrated analysis of genome-wide RNA 115 expression and DNA methylation in late onset Alzheimer’s disease and neuropathological controls. Abstract 115 Introduction 116 Results 119 Discussion 124 Methods 127 Figures 132 Tables 138 References 143 Supplemental Figures 148 Supplemental Tables 151 V. Conclusions and future directions 171 Overall summary 171 Summary of homocysteine (Hcy) and Pb exposure 172 Summary of DNA methylation in AD. 174 Summary of gene expression and DNA methylation in AD. 176 Future directions 177 Concluding remarks 181 References 182

iv LIST OF FIGURES

Chapter 1 1.1 Conceptual diagram describing the relationship between 21 environmental exposures, including to the heavy metal lead, with the development of late-onset Alzheimer’s disease. Chapter 2 2.1 One-Carbon Metabolism Pathway. Homocysteine can be 64 elevated in conditions of low folate, low B6, or low B12. The sulfhydryl groups on several proteins, including Cystathionine β- synthase (CBS), in the one-carbon metabolism pathway are potential sites for lead’s interferences. 2.2 The adjusted cross-sectional association between Pb exposure 65 and homocysteine. 2.3 Longitudinal core model. Tibia Pb. Data has been centered on 66 the mean continuous variables so intercept can be interpretable. 2.4 Adjusted association between Pb and homocysteine stratified by 67 nutrient status. S 2.1 Bivariate scatterplots for the unadjusted associations between 76 lead measurement and Hcy at visit 1. S 2.2 Blood Pb restricted plasma and diet. *Restrict analysis to only 76 individuals with both plasma and dietary measures. Chapter 3 3.1 Mean percent methylation frequency distribution of the Discovery 97 Set of 12 cognitively normal control samples (A) and 12 Alzheimer’s disease cases (B) across the 27,578 CpG sites on the Illumina HumanMethylation27 BeadArray. 3.2 Hierarchical clustering heatmap of the Discovery Set top 26 98 autosomal CpG loci associated with late-onset Alzheimer’s disease (LOAD) case/control status after adjusting for sex and age. 3.3 Discovery Set gene set enrichment analysis plots. 99 3.4 Human chromosome ideogram. 100 3.5 Methylation upstream of the TMEM59 gene. 101 3.6 Functional validation of observed DNA methylation differences for 102 TMEM59, a gene involved in the post-translational modification of Amyloid Precursor Protein (AΒPP). Chapter 4

v 4.1 Boxplots of gene expression of the 9 top genes that differ by 132 LOAD case or control status. LOAD cases display reduced gene expression at all of 9 of the genes. 4.2 Heatmap of all of the 176 sites on the Affymetrix gene expression 133 array (FDR < 0.1) associated with LOAD case status. This plot used maximum distance and Ward’s hierarchical clustering methods. Data has been normalized to the mean expression value per probeset. 4.3 Principal Component Analysis. (A) Principal component loading 134 histogram. (B). Principal component analysis scree plot for the top 176 probesets (FDR < 0.1). 4.4 Gene-expression and DNA methylation linked data analysis 135 pipeline. 4.5 (A) Scatterplot of the 133 genes that displayed discordant gene 135 expression and DNA methylation between LOAD cases and controls. 4.6 Scatter plots of 9 genes: expression vs. methylation. 136 4.7 Circos plot: Locations of gene expression change and DNA 137 methylation change. S 4.1 Heatmap of all of the 176 sites on the Affymetrix gene expression 148 array (FDR < 0.1) associated with LOAD case status. Publicly available data from embryonic stem cell derived neuronal precursor cells (GSE7178) and astrocytes (GSE5080) have been included. This plot used maximum distance and Ward’s hierarchical clustering methods and normalized to the mean expression value per probeset. S 4.2 Scatterplots of expression vs. methylation for the 23 genes 149 significantly associated with AD via expression (p<0.05) and methylation (p<0.05), and methylation and expression are significantly correlated (Pearson’s p<0.05).

vi LIST OF TABLES

Chapter 2 2.1 Univariate Statistics: Characteristics of individuals with complete Hcy and blood Pb (2234 obs, 1048 individuals). Mean (S.D.), 57 except where noted. 2.2 Bivariate statistics based on visit 1. Characteristics of individuals 5 with complete Hcy and blood Pb (1056 individuals). 58 2.3 Concurrent Pb exposure is associated with plasma 6 homocysteine: Mixed effects model, random intercept only. 60 Equivalent to cross-sectional model taking into account correlated nature of observations from the same individual. 2.4 Cumulative exposure is associated with plasma homocysteine: 6 Mixed effects model, random intercept only. Equivalent to cross- 61 sectional model taking into account correlated nature of observations from the same individual. 2.5 Longitudinal mixed effects models: Log(hcy) is outcome and tibia 6 pb is main predictor. Continuous covariates have been mean 62 adjusted so that the intercept can be interpreted as the log(hcy) at the mean of those covariates and when dummy variables =0. Four visits of tibia Pb and Hcy have been used. Random intercept and slope. 2.6 Basic mediation analysis. All visits main effect of tibia Pb with 6 and without current blood Pb adjustment. Note: Tibia Pb is no 63 longer significant after adjusting for blood Pb S 2.1 Visits 1-2 all samples, linear regression, split by change in blood 7 Pb n=747 72 S 2.2 All visits, all samples, mixed effects n=747 with 1245 7 observations 73 S 2.3 Secondary analysis, long-term dietary trend in tibia model 7 74 S 2.4 Mediation analysis, visit 1. 7 75 Chapter 3 3.1 Study population mean demographics by case status. 1 104 3.2 Pyrosequencing assay information. 1 105 3.3 Primer sequences for gene expression QPCR assays. 1 105

vii 3.4 CpG sites differentially methylated with age among cognitively 1 normal controls (Discovery Set, Age 69-94). 106 3.5 Table of the 25 CpG sites most significantly differentially 1 methylated by AD case status (Discovery Set). 107 Chapter 4 4.1 Affymetrix expression differences between LOAD cases and 1 controls (n=25) after adjusting for age and sex. 138 4.2 Among genes with lower gene expression in AD cases vs. 1 controls, the following are the top 15 biological processes that 139 are down-regulated based on LR-Path. Among genes with higher gene expression in AD cases vs. 1 controls, the following are the top 15 biological processes that 140 are up-regulated based on LR-Path. 4.4 Gene expression and methylation correlation. 1 141 4.5 Study population characteristics. 1 142 S 4.1 Gene expression probesets associated with AD case vs. controls 1 at FDR <0.1 (n=176) 151 S 4.2 Among genes with lower expression in Alzheimer’s, several 1 biological processes were enriched. 156 S 4.3 Among genes with higher expression in Alzheimer’s, several 1 biological processes were enriched. 158 S 4.4 Among genes with lower expression in Alzheimer’s, several 1 molecular functions were enriched. 166 S 4.5 Among genes with higher expression in Alzheimer’s, several 1 molecular functions were enriched. 167 S 4.6 Among genes with lower expression in Alzheimer’s, the binding 1 motifs of several transcription factors were enriched. 169 S 4.7 Among genes with higher expression in Alzheimer’s, the binding 1 motifs of several transcription factors were enriched. 169 S 4.8 Among genes with higher expression in Alzheimer’s, the binding 1 sites of several microRNAs were enriched. 169 S 4.9 Among genes with lower expression in Alzheimer’s, the binding 1 sites of several microRNAs were enriched. 169 S 4.10 Among genes with higher expression in Alzheimer’s, several 1 cytogenic bands were enriched. 170 S 4.11 Among genes with lower expression in Alzheimer’s, several 1 cytogenic bands were enriched. 170

viii LIST OF ABBREVIATIONS

8-oxo-dG 8-hydroxy-2'-deoxyguanosine AD Alzheimer's disease ADRC Alzheimer's disease research center DOHad Developmental origins of adult health and disease EOAD Early-onset Alzheimer's disease EGCG Epigallocatechin gallate GAWS Genome wide association study HCY Homocysteine KXRF Cd109 K-shell X-ray Fluorescence LOAD Late-onset Alzheimer's disease Pb Lead MCI Mild cognitive impairment NGS Next generation sequencing PND Post natal day SAM S-adenosyl methionine SNP Single nucleotide polymorphism NAS Veteran's Affairs Normative Aging Study WBC White blood cell

ix ABSTRACT The causes of sporadic neurodegenerative disease in aging adults likely involve a complex interplay of genetics, epigenetics, and a lifetime of environmental exposures. The current dissertation uses two molecular epidemiology studies to investigate biomarkers of environmental exposures and neurodegenerative outcomes.

A major challenge of chronic disease environmental etiology research is the long latency between environmental exposures and the onset of disease. The Veteran’s Affairs Normative Aging Study is an epidemiologic cohort designed to prospectively measure exposures and monitor early or subclinical disease. Repeated levels of recent exposure to lead were measured in blood and cumulative exposure to lead was measured by bone K-shell X-ray fluorescence. Homocysteine (Hcy), a risk factor for cardiovascular and neurodegenerative diseases when elevated, was measured concurrently with blood lead. Using repeated measures mixed effects models, this research demonstrated that an interquartile range higher blood Pb level (3 µg/dl) was associated with an 8.1% higher Hcy, compared to the percent change in Hcy with a 5-year increase in age (3.1%). We also demonstrated that individuals with diets rich in vitamins B6, B9, and B12, mitigated the effect of Pb on Hcy. This research suggests that interventions to reduce blood Pb and increase dietary B-vitamin intake would reduce circulating Hcy levels, potentially lowering risk for cardiovascular and neurodegenerative disease.

The second study investigated the role of epigenetic regulation, through DNA methylation, and its potential contribution to gene expression changes in late-onset Alzheimer’s disease. In two separate thesis papers, the DNA methylomes and transcriptomes of frontal cortex tissues from deceased patients with Alzheimer’s disease were mapped and compared to neuropathologically

x normal controls using genome-wide approaches and bioinformatic analyses. In a proof-of-concept study, the top disease ranked DNA methylation site was validated for altered gene expression and protein levels. A novel biomarker and potential mechanism for LOAD pathogenicity with environmental implications was proposed.

This interdisciplinary thesis implemented laboratory biomarker studies, population exposure assessment and molecular epidemiology to approach the multi-faceted origins of neurological disease in aging populations.

xi CHAPTER I Introduction

Alzheimer’s Disease and Environmental Exposure to Lead: The Epidemiologic Evidence and Potential Role of Epigenetics

From: Bakulski KM, Rozek LS, Dolinoy DC, Paulson HL, Hu H. 2012. Alzheimer’s disease and environmental exposure to lead: The epidemiologic evidence and potential role of epigenetics. Current Alzheimer’s Research 9: 563-73, reprinted with permission for educational use from Bentham Science Press.

ABSTRACT

Several lines of evidence indicate that the etiology of late-onset Alzheimer’s disease (LOAD) is complex, with significant contributions from both genes and environmental factors. Recent research suggests the importance of epigenetic mechanisms in defining the relationship between environmental exposures and LOAD. In epidemiologic studies of adults, cumulative lifetime lead (Pb) exposure has been associated with accelerated declines in cognition. In addition, research in animal models suggests a causal association between Pb exposure during early life, epigenetics, and LOAD. There are multiple challenges to human epidemiologic research evaluating the relationship between epigenetics, LOAD, and Pb exposure. Epidemiologic studies are not well-suited to accommodate the long latency period between exposures during early life and onset of Alzheimer’s disease. There is also a lack of validated circulating epigenetics biomarkers and retrospective biomarkers of Pb exposure. Members of our research group have shown bone Pb is an accurate measurement of historical Pb exposure in adults, offering an avenue for future epidemiologic studies. However, this would not address the risk of LOAD attributable to early-life Pb exposures. Future studies

1 that use a cohort design to measure both Pb exposure and validated epigenetic biomarkers of LOAD will be useful to clarify this important relationship.

Keywords

DNA methylation; epigenetics; epidemiology, Late-onset Alzheimer’s disease; lead exposure; Pb

ALZHEIMER’S DISEASE

General Alzheimer’s Disease Epidemiology

Alzheimer’s disease (AD) is a highly prevalent, progressive, and fatal neurodegenerative disease associated with aging. Clinical manifestation of AD includes progressive memory impairment and a gradual difficulty performing normal activities. A small percentage of cases, termed early-onset AD (EOAD), experience disease onset prior to age 60. EOAD cases are attributed to highly penetrant genetic mutations in amyloid pathway genes including amyloid precursor protein (APP) on chromosome 21, presenilin 1 (PSEN1) on chromosome 14, and presenilin 2 (PSEN2) on chromosome 1 (Bertram 2009; Hardy 1997). These mutations lead to the accumulation of β-amyloid plaques, a pathological hallmark of AD.

Termed late-onset AD (LOAD), the majority of AD cases are sporadic and symptoms manifest after age 60. Numerous low-penetrant genetic risk factors conferring a modest increase in risk of disease have been identified for LOAD, the most studied of which is the apolipoprotein ε4 allele (APOE-ε4). The global population prevalence of APOE-ε4 is 22%, while approximately 60% of LOAD cases carry at least one allele (Ashford 2004; Kim et al. 2009). Large, multi- center genome-wide association studies (GWAS) estimate the population attributable risk for APOE variants is 19-35% (Ertekin-Taner 2010). GWAS have

2 identified additional polymorphisms associated with LOAD risk including genes for ABCA7, BIN1, CD2AP, CD33, CLU, CR1, EPHA1, MS4A, and PICALM (Harold et al. 2009; Hollingworth et al. 2011; Lambert et al. 2009; Naj et al. 2011), each associated with small increases in population attributable risk (PAR) ranging from 2-9.3% with a combined non-APOE PAR of 31-35%. Additional APOE ε4 dose adjustment reveals 50% of the PAR for LOAD is accounted for by known single nucleotide polymorphisms (SNPs) (Naj et al. 2011). While these variants are important both for risk assessment and identification of novel mechanisms of pathogenesis, they are neither necessary nor sufficient for the development of LOAD.

Twin studies are an important epidemiologic tool for estimating the relative contribution of genetics and the environment in disease development. Incomplete twin concordance and variable age of onset supports a significant role for non-genetic factors in LOAD etiology. Among monozygotic twin (MZ) pairs, approximately 45-67% of twin pairs are concordant for LOAD (Gatz et al. 1997; Gatz et al. 2006; Nee and Lippa 1999). Heritability of liability based on twin studies is estimated to be 58-79% (Gatz et al. 1997; Gatz et al. 2006). Linkage analysis reveals age at LOAD onset is partially genetically linked to regions on chromosomes 4 (208 cM) and 10 (139 cM) (Li et al. 2002). However, among a group of MZ pairs in which both twins develop the disease, differences in age of onset range from 4 to 16 years (Gatz et al. 1997). Both genetic and environmental factors likely contribute to LOAD development.

Association studies have identified several non-genetic risk factors for LOAD, including depression (Saczynski et al. 2010), hypertension (Li et al. 2011; Sharp et al. 2011), stroke (Savva and Stephan 2010), diabetes (Li et al. 2011), hypercholesterolemia (Li et al. 2011), obesity (Anstey et al. 2011), head trauma (Plassman et al. 2000), smoking (Almeida et al. 2002; Li et al. 2011; Rusanen et al. 2011), and having greater than 6 siblings (Moceri et al. 2000). Protective factors, those that reduce the risk of developing LOAD or delay the onset of LOAD, include physical activity (Laurin et al. 2001; Lindsay et al. 2002), social

3 engagement (Fratiglioni and Wang 2007), mental activity (Fratiglioni and Wang 2007; Stern and Munn 2010), education (via the cognitive reserve hypothesis) (Fratiglioni and Wang 2007; Lindsay et al. 2002), statin use (G Li et al. 2010) , non-steroidal anti-inflammatory drug (NSAID) use (Lindsay et al. 2002), moderate alcohol consumption (Larrieu et al. 2004; Lindsay et al. 2002), coffee consumption (Lindsay et al. 2002), past vaccinations (Verreault et al. 2001), and childhood residence in the suburbs relative to the city (Moceri et al. 2000). In particular, nutrition may play a protective role in LOAD onset. Consumption of one meal/week of fish rich in omega-3 fatty acids reduced the risk of developing AD by 60% in the Chicago Health and Aging Project (Morris 2009). Individuals with plasma vitamin E less than or equal to 21.0 µmol/L had a higher risk of incident dementia than individuals with plasma levels greater than or equal to 25.5 µmol/L (Helmer et al. 2003). The natural plant polyphenols curcumin and green tea epigallocatechin gallate (EGCG) have anti-oxidant and neuroprotective properties that may be protective against LOAD (Mandel et al. 2007). EGCG reduces APP translation through modulation of the intracellular iron pool in vitro neuroblastoma cell culture (Reznichenko et al. 2006) and AD transgenic mice exposed in vivo to EGCG show reduced Aβ plaque density (Rezai-Zadeh et al. 2005). In an additional AD transgenic mouse model study, curcumin suppressed inflammation and oxidative damage in the brain and lowered levels of soluble Aβ and plaques (Yang et al. 2005). Anthropometric measures of shorter adult knee height and arm span may reflect nutritional deficits in childhood (Jeong et al. 2005) and women in the lowest quartile of arm span in the Cardiovascular Health Cognition cohort study had 1.5 times elevated risk of dementia (Huang et al. 2008).

Proposed environmental exposures associated with LOAD include aluminum (Frisardi et al. 2010; Shcherbatykh and Carpenter 2007), copper (Brown 2009; Shcherbatykh and Carpenter 2007), zinc (Shcherbatykh and Carpenter 2007), mercury (Gerhardsson et al. 2008), lead (reviewed below), iron (Mandel et al. 2007), pesticides (Baldi et al. 2003; Santibanez et al. 2007), solvents (Kukull et al. 1995), electromagnetic field (Sobel et al. 1996), and

4 particulate matter in air pollution (Calderon-Garciduenas et al. 2004). Environmental exposure studies have been underrepresented in the AD literature, likely due to the challenges of retrospective exposure assessment in older adults.

LEAD (Pb) EXPOSURE

Overview of Pb Exposure as a Risk Factor

Zawia and colleagues have published a series of experimental studies on rodents and primates demonstrating that Pb exposure in early life results in late- life neuropathological changes similar to those of AD (reviewed elsewhere in this issue of Current Alzheimer’s Research). This work, coupled with the recognition that exposure to Pb in the general population until recently has been high, has heightened interest in the epidemiology of Pb exposure and neurodegenerative disease. We discuss trends in Pb exposure and epidemiologic studies that provide evidence for the role of Pb as a risk factor for AD.

Pb Exposure Epidemiology

One of the greatest environmental health successes of our society was the regulatory action to reduce what had been decades of high Pb exposure in the US (Grosse et al. 2002). Between 1976 and 1991, the mean blood Pb levels for people in the US dropped 78% from 12.8 µg/dL to 2.8 (Pirkle et al. 1994). Since 1991 the standard elevated blood Pb level defining the need for action from Pb poisoning in children has been set to 10 µg/dL. The previously elevated mean blood Pb level of 12.8 µg/dL is a sobering testament to the high levels of Pb exposure endured by the general US population and other countries in the recent past.

However, hazardous public health impacts remain despite low mean population blood Pb levels. One issue is that the general reduction in Pb exposure is not universal. Pockets of high Pb exposure remain in certain sectors

5 of the US population where housing was constructed prior to 1950 and at which time leaded paint was used (Lanphear et al. 1998; Rabito et al. 2003), or where plumbing pipes and solder containing Pb have not yet been replaced (Edwards et al. 2009). Fly ash from municipal waste incineration contains high levels of heavy metals including Pb (Jung et al. 2004). In the high temperatures of the trash incineration process, Pb is converted to the volatile PbCl2 compound, which can contaminate surrounding areas (Jung et al. 2004). It is estimated that the US has hundreds of defunct Pb battery recycling sites (Nedwed and Clifford 1997) and Pb/zinc mines and smelters. The surrounding soil and water of these industrial sites are often contaminated with high levels of Pb, leading to human exposure (Kaul et al. 1999), and many of these sites are designated Superfund sites on the National Priorities List by the US Environmental Protection Agency (EPA) (Lewin et al. 1999; von Lindern et al. 2003). In many developing countries the combustion of leaded gasoline continues and industrial emissions of Pb have been increasing (Meyer et al. 2008; Tong et al. 2000). Groups of people continue to experience high blood Pb levels based on their occupation or residential proximity to these hazards.

The pharmacodynamics of Pb in the human body makes past exposure to this heavy metal relevant to current and future health outcomes. Pb dust is inhaled or ingested and absorbed through the lung epithelia or gastrointestinal tract respectively. Pb is taken up by divalent metal transporters in the gut, binds tightly to heme molecules, and circulates throughout the body via blood. A small percentage of circulating Pb is highly toxic because it is free and bioavailable in the plasma (Hernandez-Avila et al. 1998). Plasma Pb contributes to both soft tissue Pb as well as bone Pb deposition (Rabinowitz et al. 1976). Pb can occupy both Ca2+ sites in the hydroxyapatite structure of bones (Barry and Mossman 1970) and greater than 95% of the adult body burden of Pb is stored in bones (Barry and Mossman 1970). Given that cortical bone turns over at a slow rate of approximately 2% per year in healthy adults, Pb can be stored for decades in bone (Barbosa et al. 2005; Hu et al. 1998; Rabinowitz 1991). Storage in bone is not a permanent Pb detoxification mechanism as Pb can have direct effects on

6 the cellular components of bone (Pounds et al. 1991), and bone Pb can be mobilized in times of higher bone turnover such as during pregnancy, lactation, and osteoporosis (Silbergeld et al. 1988). Individuals born in the US prior to the Pb phase out in the 1970s may have accumulated elevated bone Pb stores that become mobilized in later life.

Biomarkers of Pb Exposure

Whole blood Pb is the most common biomarker of Pb exposure. The half- life of Pb in blood is relatively short, approximately 35 days (Rabinowitz et al. 1976). This biomarker is best used for quantifying recent environmental exposures and mobilization of endogenous Pb (Silbergeld 1991). Similarly, soft tissues also have relatively high turnover of Pb with a mean half-life of approximately 40 days, but soft tissue Pb quantification is invasive and not typically used for epidemiologic studies (Rabinowitz et al. 1976).

An expert panel on adult Pb toxicity convened by the US Centers for Disease Control concluded that bone Pb levels were the best biomarker of cumulative Pb exposure (Hu et al. 2007). Spongy trabecular bone, such as that found in the patella, has an intermediate half-life of 5-15 years in adults (Hu et al. 1995). More dense cortical bone, as in the tibia, has a much longer half-life of 10-30 years (Chettle 2005). Thus, epidemiologic studies measuring Pb in bone can quantify a subject’s life history of cumulative Pb exposure.

Bone Pb levels can be measured either in vivo using Cd109 K-shell X-ray Fluorescence (KXRF) (Hu et al. 2007) or by direct, chemical measurement of Pb in excised total joint replacement or post-mortem bone samples (Wittmers et al. 1988). Measurements with KXRF technology are painless and non-invasive, with minimal radiation exposure (Hu et al. 1995). The KXRF instrument uses low- level gamma radiation to provoke emission of fluorescent photons from a subject’s tibia and patella (Hu et al. 1989). The photons are detected and quantified over a spectrum of wavelengths from which the characteristic emission profile of Pb can be extracted (Hu et al. 1995). Post-mortem bone samples can

7 be acid digested and quantified for Pb levels using inductively coupled plasma mass spectrometry (ICP-MS) (Garcia et al. 2001).

There are several approaches to predicting cumulative Pb exposure in the absence of direct bone Pb measures. The Park model incorporates blood Pb and information on subject demographics, medical history, and metabolic parameters to predict cumulative Pb exposure (Park et al. 2009). The Gorell system predicts cumulative Pb exposure based on blood Pb levels and physiologically based pharmacokinetic (PBPK) models incorporating industrial hygienist rated occupational Pb exposure for each job over the duration worked (Coon et al. 2006). Both strategies have been validated with bone Pb measures.

Pb as a Neurotoxicant and Risk Factor for LOAD

Pb is a well-known neurotoxicant in children. Even at relatively low (subclinical) levels, epidemiologic studies demonstrate that childhood Pb exposure affects IQ and behavior with major impacts on IQ and functioning (Fewtrell et al. 2004; Grosse et al. 2002).

A growing body of toxicological and population-based research indicates that cumulative environmental Pb exposure is neurotoxic in adults as well (Toscano and Guilarte 2005). Pb exposure is a significant risk factor for accelerated declines in cognition (Weisskopf et al. 2004; Wright et al. 2003), an effect that a recent CDC panel concluded was likely causal (Shih et al. 2007). The Veteran’s Affairs Normative Aging Study (NAS) is a longitudinal cohort of men free of disease when recruited in 1963. Based on data from repeated measures of bone Pb, blood Pb, and cognitive tests in the NAS, there are significant associations between high Pb exposure and decreased cognition. The cognitive domains associated with increased Pb exposure differ depending on the time of exposure. In a cross-sectional analysis, higher blood Pb was associated with reduced ability to recall and define words, identify line-drawn objects, and difficulty with a perceptual comparison test (Payton et al. 1998). Both higher blood and bone Pb were associated with decreased spatial copying

8 skill (Payton et al. 1998). Higher bone Pb was associated with reduced pattern memory (Payton et al. 1998). Longitudinal analyses confirm that the Pb associated declines in cognitive function are greater than changes observed with normal aging alone (Schwartz et al. 2000). Also in the NAS, functional genetic polymorphisms in the δ-aminolevulinic acid dehydratase (ALAD) and hemochromatosis (HFE) genes modify the association between Pb and cognition measured by the Mini-Mental State Exam (MMSE), where variant carriers have more pronounced cognitive deficits associated with Pb exposure (Wang et al. 2007; Weuve et al. 2006). Future research may study the ecological association between geographic regions with elevated Pb exposure and prevalence of LOAD.

Pb is also a risk factor for increased hippocampal gliosis measured by magnetic resonance spectroscopy in the NAS (Weisskopf et al. 2007), an abnormality associated with LOAD development. Molecular epidemiology studies show cumulative Pb exposure is associated with an increased risk of amyotrophic lateral sclerosis (Kamel et al. 2005; Kamel et al. 2003; Kamel et al. 2002) and Parkinson’s Disease (Weisskopf et al. 2010), suggesting that Pb exerts a significant neurodegenerative effect. This effect may have specificity through epigenetic change as a pathogenic mechanism.

Toxicological studies are consistent with the epidemiologic research. Early life Pb exposure in animal models is associated with latent APP pathway dysregulation. Rats exposed to Pb in early life showed increased expression of APP mRNA and elevated Aβ aggregation without changes in α-, β-, or γ- secretases at 20 months of age (Basha et al. 2005a; Basha et al. 2005b). Similarly, primates exposed to Pb during the first 2 months of life only had significant adverse brain changes at 23 years of age when compared to their unexposed counterparts (Wu et al. 2008a). Pb exposed primates had increased amyloidogenesis, senile plaque deposition, and up-regulation of key proteins in the amyloid processing pathway, such as APP and beta-site APP-cleaving enzyme 1 (BACE1) (Wu et al. 2008a).

9 OVERVIEW OF EPIGENETICS

Literally meaning “above the genome,” the epigenome comprises the heritable changes in gene expression that occur in the absence of changes to the DNA sequence itself. Epigenetic mechanisms include chromatin folding and attachment to the nuclear matrix, packaging of DNA around nucleosomes, covalent modifications of histone tails, and DNA methylation. The influence of regulatory small RNAs and micro RNAs on gene transcription is also increasingly recognized as a key mechanism of epigenetic gene regulation (Morris 2011). Epigenetic mechanisms are important in growth and cellular differentiation (Jones and Taylor 1980). Epigenetic change can be stochastic (Feinberg and Irizarry 2010) or internally orchestrated as part of aging (Fraga and Esteller 2007). Longitudinal change in global and gene-specific DNA methylation clusters within families, suggesting that there is genetic control of methylation status (Bjornsson et al. 2008). Inappropriate epigenetic changes are associated with many diseases including cancers (Esteller 2008), Rett syndrome (Horike et al. 2005), Beckwith-Wiedemann syndrome (DeBaun et al. 2002) and other imprinting disorders. Environmental signals can trigger epigenetic responses and may be an important mechanism by which environmental exposures are associated with disease (Faulk and Dolinoy 2011). Furthermore, epigenetic mechanisms may play an important role in the developmental origins of adult health and disease (DOHaD) by providing a mechanism underlying the latent effects of adverse fetal, infant, and childhood environments on late-life chronic disease (Barker 2004; Hanson et al. 2011; Wadhwa et al. 2009).

Epigenetic Epidemiology and Alzheimer’s Disease

Epigenetic epidemiology is the study of the effects of heritable epigenetic changes on the occurrence and distribution of diseases in populations (Jablonka 2004). This research includes both trans-generational and intra-individual cellular epigenetic inheritance systems. Epigenetic changes are associated with epidemiologic risk factors such as aging (Calvanese et al. 2009; Fraga 2009) and environmental exposures (Faulk and Dolinoy 2011), as well as psychiatric

10 outcomes (Sananbenesi and Fischer 2009) and neurodegeneration (Urdinguio et al. 2009).

Evidence for the role of epigenetics in AD pathogenesis is found in human studies of various tissues, animal models, and cell culture (2010; Mastroeni et al. 2011; Mill 2011). Global changes associated with AD have been observed in DNA methylation, miRNAs, and histone modifications. A human post-mortem case-control study identified global DNA hypomethylation in the entorhinal cortex of AD subjects by quantifying the percentage of positive 5-methylcytosine neuronal nuclear immunoreactivity (Mastroeni et al. 2008). Within a single MZ twin pair discordant for AD, DNA from the temporal neocortex neuronal nuclei was hypomethylated in the AD twin compared to their cognitively normal twin using similar methods to the previous study (Mastroeni et al. 2009). An AD case- control study in the post-mortem human parietal lobe cortex revealed differential regulation of miRNAs including miR-204, miR-211, and miR-44691 using a custom µParaflo array (Nunez-Iglesias et al. 2010). Age-matched AD cases have increased neuronal global phosphorylation of histone 3 relative to controls determined by immunolabeling in the hippocampus, a histone modification that suggests mitotic activation (Ogawa et al. 2003).

Given that epigenetics play an important regulatory role in gene expression, epigenetic dysregulation of important AD tau and amyloid processing pathway genes may point to a potential mechanism for AD disease progression. In experiments where neuroblastoma cells were cultured under low folate and vitamin B12 conditions, PSEN1 and BACE1 were hypomethylated, mRNA expression of BACE1 and PSEN1 was significantly induced, and Aβ production was increased (Fuso et al. 2005). Addition of S-adenosyl methionine (SAM) was able to restore BACE1 and PSEN1 expression to baseline levels, though DNA methylation reversal was incomplete (Fuso et al. 2005). An additional study using human neuroblastoma cells and male rat brain tissue shows APP mRNA expression is repressed by thyroid hormone (T3) sensitive histone modifications (Belakavadi et al. 2011). Treatment with T3 decreases H3K4 methylation and H3

11 acetylation at the APP promoter, leading to APP silencing that was reversed with histone deacetylase (HDAC) and histone lysine demethylase inhibitors (Belakavadi et al. 2011).

There have been several candidate-gene methylation studies in LOAD cases and controls. In a post-mortem brain study of 26 controls and 44 LOAD cases with varying degrees of disease severity, no differences were seen in DNA methylation in regions associated with Microtubule Associated Protein Tau (MAPT), PSEN1, and APP, nor were differences detected between frontal cortex and hippocampal DNA (Barrachina and Ferrer 2009). Investigation of 6 familial AD frontal cortex and cerebellum brain samples revealed no methylation at the APP promoter in any case in either brain region (Brohede et al. 2010). These studies were limited by a candidate-gene approach and highlight the need for genome-wide assessment of DNA methylation.

It is critical that epigenetic epidemiology studies of AD epigenetics consider age as an independent predictor of epigenetic change as age-specific epigenetic drift has been observed at AD related loci among healthy normal controls. In a set of control parietal cortex samples, the promoter of APP was hypomethylated in individuals greater than 70 years of age relative to younger subjects (Tohgi et al. 1999a). DNA methylation upstream of the MAPT gene also varied with age in the control parietal cortex and was associated with an age- related decline in MAPT gene expression (Tohgi et al. 1999b). Specifically, MAPT promoter CpG dinucleotides located in the Sp1 transcriptional activator binding site were hypermethylated with age, while CpG dinucleotides located within the GCF transcriptional repressor binding region were hypomethylated with age (Tohgi et al. 1999b). Another study of post-mortem cerebral cortex in 125 subjects ranging from 17 weeks of gestation to 104 years of age measured methylation by MethyLight PCR at candidate tag loci for 50 genes selected for their relevance to LOAD, CNS differentiation, and cancer. CpG sites in the promoters of eight genes showed robust linear increases in DNA methylation across the lifespan (Siegmund et al. 2007). An additional study examined

12 prefrontal cortex samples across a 30 year age range and noted that the average DNA methylation in promoters of MTHFR and APOE increased by 6.8% across the age range, while control samples decreased by 10.6% with age (Wang et al. 2008). Given the likely epigenetic drift, clinical samples should be carefully matched on age.

Epigenetics and Heavy Metals, with a Focus on Pb

Epigenetic alterations have been observed following exposure to environmental metals (Salnikow and Zhitkovich 2008), including arsenic, nickel, chromium, cadmium, and Pb. Perhaps the heavy metal most studied in the field of cancer epigenetic epidemiology is arsenic. In a population-based study of 351 individuals with bladder cancer, elevated toenail arsenic measurements were associated with increased tumor sample promoter methylation of RASSF1A and PRSS3 tumor suppressor genes (Marsit et al. 2006). Nickel, chromium, and cadmium epigenetics research has largely been in toxicologically based in vitro experiments. A cell line of human lung bronchoepithelial cells treated with nickel chloride show global histone modification changes including decreased H2A, H2B, H3, and H4 acetylation and increased H3K9 dimethylation (Ke et al. 2006). When the same cell line is treated with chromium, the cells exhibit increased H3K9 dimethylation at the MLH1 gene promoter region, which correlates with decreased MLH1 mRNA expression (Sun et al. 2009). Cadmium exposure in a rat liver cell line initially reduces DNA methyltransferase activity and global DNA methylation, but after 10 weeks of prolonged exposure, the cells show significant increases in DNA methyltransferase and global DNA methylation above the baseline (Takiguchi et al. 2003).

Evidence suggests that Pb, in particular, may play a role in epigenetics throughout the life course. In a study of 103 mother-infant pairs, maternal cumulative Pb exposure was inversely associated with offspring umbilical cord genomic DNA methylation of Alu retrotransposable elements (Pilsner et al. 2009). Similarly, bone Pb levels were inversely associated with peripheral blood genomic DNA methylation of LINE-1 retrotransposons in 517 elderly men from

13 the NAS (Wright et al. 2010). Individuals exposed to extremely high levels of Pb (51-100 µg/dL blood Pb) had higher methylation in the promoter of the p16 tumor suppressor gene (Kovatsi et al. 2010). Research is needed to expand this early epidemiologic work on global and candidate gene DNA methylation to more comprehensively understand specific pathways influenced by Pb exposure in humans.

Animal studies have investigated the relationship between Pb exposure and epigenetics. Early life exposure to Pb in primates causes dysregulation of biological pathways important to LOAD pathogenesis in late life and is associated with reduced DNA methyltransferase 1 (DNMT1) activity (Wu et al. 2008a). Rat pheochromocytoma cells exposed to Pb show dose dependent decreases in global methylation and decreases in APP promoter methylation at 4 CpG sites (YY Li et al. 2010). These changes were associated with increases in APP mRNA and Aβ protein levels (YY Li et al. 2010). Toxicological and epidemiological studies suggest that Pb exposure may be associated with epigenetic change, but further research is needed.

DATA INTEGRATING ALZHEIMER’S DISEASE, EPIGENETICS, AND Pb EXPOSURE

Alzheimer’s and Pb Exposure are Associated with Changes in One-Carbon Metabolism, the Substrate for DNA Methylation

De novo and maintenance DNA methylation is dependent on available methyl (-CH3) groups. One-carbon metabolism reactions are reversible and deficiencies in methyl donors can cause DNA hypomethylation. For example, mice given diets deficient in the methyl donor choline showed lower global brain methylation (Niculescu et al. 2006) and elevated expression of APP, consistent with promoter hypomethylation (Niculescu et al. 2005). Epidemiologic studies indicate that AD patients have altered circulating levels of one-carbon metabolism members including homocysteine (HCY), SAM, folate, and vitamin B12. Elevated HCY is associated with increased risk of developing AD and

14 increased rate of disease progression among individuals with the disease. Prospective data from the Framingham Heart Study show that each standard deviation increase in log transformed plasma total HCY levels was associated with an adjusted relative risk of dementia of 1.8 (95% CI: 1.3-2.5) eight years after the HCY measurement (Seshadri et al. 2002). AD patients in the Oxford Project to Investigate Memory and Ageing have increased serum HCY relative to cognitively normal control subjects (n=164) and the individuals with the greatest disease progression over the subsequent three years had the highest original HCY levels (Clarke et al. 1998). SAM is a methyl-donor molecule that is hydrolyzed to form HCY, the substrate for DNA methylation. AD patients also have decreased cerebrospinal fluid SAM relative to cognitively normal controls (Bottiglieri et al. 1990).

Several proteins in the one-carbon metabolism cycle may be disturbed by Pb exposure because elemental Pb reacts with free sulfhydryl groups on proteins. HCY metabolism may be directly inhibited by Pb binding to the sulfhydryl group in HCY. Furthermore, HCY is transsulfurated into cysteine by cystathionine β-synthase (CBS) and CBS has two sulfhydryl groups with which Pb can react. There is also evidence for Pb’s involvement in methionine processing. Rats developmentally treated with Pb have impaired long-term potentiation (LTP), memory, and synaptic plasticity. Co-treatment with SAM and Pb increases LTP relative to Pb treatment alone and reduces circulating blood Pb levels (Cao et al. 2008) . Similarly, neuroblastoma cells exposed to Pb experience viability loss, glutathione antioxidant depletion, membrane lipid peroxidation, DNA damage, and apoptosis; pretreatment with a methionine derivative reduces these harmful effects (Chen et al. 2011).

Pb exposure and HCY levels are linked in cross-sectional epidemiologic studies. In the Baltimore Memory and Aging Project involving greater than 1,000 adults, higher blood Pb was associated with higher HCY (Schafer et al. 2005). Analyses from the 1999-2002 National Health and Nutrition Examination Survey (NHANES) showed HCY was strongly associated (OR=1.92) with peripheral

15 arterial disease (PAD) (Guallar et al. 2006). Subsequent analysis showed the original association was actually due to confounding from smoking, blood Pb and cadmium levels, and impaired renal function (Guallar et al. 2006). This suggests that the association between HCY and chronic disease may be driven by environmental exposures.

Animal Model Studies Linking Pb Exposure, Epigenetics, and Amyloidogenesis

A series of rat and primate model studies conducted by the Zawia research group collectively demonstrate that early life Pb exposure reduces DNA methyltransferase activity and specifically alters the regulation of many AD pathway related genes including APP and BACE1 that are known to be CpG rich. Rats exposed to Pb from post natal day (PND) 1 through PND 20 experienced a transient increase in APP mRNA expression in cortical brain tissue, which returned to basal levels at 1 year, and later resurged at 20 months of age in the absence of continued exposure (Basha et al. 2005b). The observed late-life rise in APP mRNA was accompanied by elevated Aβ, suggesting that early life Pb exposure may have long-term effects on amyloidogenesis in late life (Basha et al. 2005b). In a follow-up study on the same tissues, investigators noted the effects on Aβ formation and aggregations were not due to changes in protein levels of APP processing secretases (Basha et al. 2005a). In a third study using the early-life exposed rat brain tissues, elevated oxidative DNA damage measured by cerebral 8-hydroxy-2’-deoxyguanosine (8-oxo-dG) was observed in the exposed animals (Bolin et al. 2006). Local 8-oxo-dG is associated with hypomethylation at adjacent CpG sites (Cerda and Weitzman 1997). Direct oxidation of 5-methylcytosine to 5-hydroxymethylcytosine may be part of active DNA demethylation (Wu and Zhang 2010). Analogous primate experiments by Zawia et al. are consistent with these rodent findings. Primates exposed in early life to Pb had elevated levels of the A peptide, 8-oxo-dG DNA, and mRNA from APP and BACE1 on autopsy 23 years later relative to controls, suggesting Pb is involved in LOAD-like pathology (Wu et al. 2008a). Brain tissue from these

16 exposed primates also had 20% reduced DNA methyltransferase 1 activity (Wu et al. 2008a) and lower methylation at the promoter of APP (Wu et al. 2008b). In vivo animal model studies spanning multiple organisms support an integrated role of Pb exposure and epigenetics in amyloidogenesis.

CHALLENGES TO LOAD EPIDEMIOLOGIC RESEARCH INTEGRATING EPIGENETICS AND Pb EXPOSURE

Human epidemiologic research integrating LOAD, environmental exposure to Pb, and epigenetics faces many challenges. Clinical criteria for AD include progressive impairment in memory in the absence of motor, sensory, or coordination deficits (McKhann et al. 1984). However, the standard of diagnosis for AD requires the pathologic post-mortem identification of Aβ plaques and tau neurofibrillary tangles. Epidemiologic studies can take advantage of predictive and diagnostic biomarkers, including a panel of plasma signaling proteins (Ray et al. 2007), cerebrospinal fluid protein analyses (De Meyer et al. 2010), magnetic resonance imaging (MRI) volumetric and structural measures (Jack et al. 1992), and positron emission tomography (PET) neuroimaging of metabolic rate and Aβ pathology (Klunk et al. 2004; Minoshima et al. 1995). However, these research methods require additional validation to become routine early detection methods (Dubois et al. 2007).

Another concern in environmental epidemiology is that the length of time between exposure and disease onset. Barker first introduced the hypothesis that early-life conditions could be linked to late life chronic disease, otherwise known as the developmental origins of health and disease (DOHaD) hypothesis (Barker and Osmond 1986). Fetal or childhood exposures have been associated with adverse health outcomes including impaired glucose tolerance (Ravelli et al. 1998) and hypertension (Barker et al. 1990; Bergvall et al. 2007). Indeed, several early life events related to growth, metabolism, and cognitive reserve have been associated with LOAD (Miller and O'Callaghan 2008). AD risk is

17 increased with limited education and income, and both factors are associated with poor early life environment and growth (Borenstein et al. 2006). Middle life risk factors including obesity (Whitmer et al. 2005), limited physical activity (Rovio et al. 2005), and diabetes (Luchsinger et al. 2001) are shared between AD and cardiovascular disease. Low birth weight and intrauterine growth restriction are related to metabolism, fat distribution, and insulin resistance at mid-life and it has been suggested that these early-life events may be associated with AD as well (Landrigan et al. 2005; Lester-Coll et al. 2006; Ross et al. 2007). However, LOAD is a chronic disease of old age and a prospective developmental exposure study could not feasibly follow an early life cohort for 75 years with our current late stage diagnostic measures. Additionally, retrospective exposure assessment is difficult. The human body has efficient detoxification and clearance mechanisms for many toxicants and many chemicals do not bioaccumulate in the human body. There is an acute need to develop biomarkers that correspond to prior toxicologic exposures.

Finally, an additional roadblock is that brain specific epigenetic measurements are only possible post-mortem. Molecular epidemiology research of toxicant induced disease is strengthened when performed with relevant tissue samples. Brain tissue collection is invasive and not possible longitudinally on live subjects. Model animal research and epidemiology studies of human pre- mortem available tissues such as skin, blood, colon, etc. are necessary to fill in stages of disease tissue not available through end of life epidemiologic brain banks.

POTENTIAL APPROACHES TO STUDY Pb EXPOSURE, EPIGENOMICS, AND ALZHEIMER’S DISEASE EPIDEMIOLOGY

To best understand the relationship between Pb exposure (both early-life and later life) and LOAD, studies should take advantage of available biomarkers of Pb and technologic advances in epigenetic measurements. Bone Pb levels are a strong predictor of negative health outcomes including elevated risks for hypertension (Cheng et al. 2001; Hu et al. 1996; Korrick et al. 1999), ischemic

18 heart disease (Jain et al. 2007), and mortality (Weisskopf et al. 2009), but the relationship between cumulative Pb exposure and LOAD has not been assessed. Cumulative Pb exposure of LOAD subjects can be measured either non- invasively in vivo using K-x-ray fluorescence (Hu et al. 2007) or by direct measurement of Pb in bone samples (Wittmers et al. 1988). At Alzheimer’s Disease Research Centers (ADRCs) where LOAD subjects consent to brain tissue donation on autopsy, it would be most ideal to directly measure Pb in samples of cranial bone obtained at the time of brain harvesting. Measurement of Pb in the cranium is highly correlated with a weighted average of skeletal Pb levels, as well as the level of Pb in tibia bone (Hu et al. 1990), the latter being the bone most commonly measured in epidemiologic studies of chronic Pb toxicity (Hu et al. 2007). Sampling cranial bone Pb would make it possible to concurrently study LOAD epidemiology, brain tissue epigenetics and cumulative Pb exposure in post-mortem case-control studies.

Circulating epigenetic biomarkers would be useful to conduct case-control studies of Pb exposure (by in vivo KXRF) with live subjects. Post-mortem Alzheimer’s disease brain tissue epigenetic studies are expanding, but use of this tissue collected at end of life is not feasible to track within individual changes over time as in longitudinal epidemiological aging cohort studies. Biologically- available biomarkers would allow for repeated epigenetic measures throughout the disease course. Epigenetic markers in white blood cells (WBC) have been used as biomarkers in other diseases. Global DNA methylation has been associated with several cancers, myelodysplastic syndrome, and schizophrenia and thus does not appear to be a disease specific biomarker. Gene-specific methylation data and risk factor methylation data are more limited and results are inconsistent (Terry et al. 2011). Larger, prospective cohort studies are needed to determine whether WBC gene-specific epigenetics will be informative with AD and with Pb exposure (Figure 1.1). Upon epigenetic biomarker development, cohort studies could integrate and target distinct age groups. Birth cohorts could investigate the role of in utero and postnatal Pb exposure on AD biomarkers to test the hypothesis suggested by animal research (Wu et al. 2008b) that early life

19 is a critical window for Pb’s influence on developmental reprogramming. Mid-life cohorts could focus on later exposure periods and could incorporate traditional AD risk factors such as hypertension status and education achieved. Late-life cohorts would involve the best AD and mild cognitive impairment (MCI) diagnostic tools and study the role of cumulative lifetime Pb exposure.

Finally, epidemiologic data needs to be incorporated with epigenetic studies on ADRC brain bank tissues. Epigenetic changes are associated with age (Wang et al. 2008), sex (Anway et al. 2005), exposures (Dolinoy et al. 2007), and diseases (Waterland and Garza 1999). Alzheimer’s disease specific epigenetic change may need to be extracted from a noisy background of age- specific epigenetic drift, sex-specific epigenetic marks, co-morbidity disease changes, and a lifetime of environmental exposures. The majority of existing studies of brain epigenetics focus on CpG islands and the application of array- based approaches that only cover a portion of the genome, largely in genic regions. Rapid advances in technology and reduction in costs have made new approaches using next-generation sequencing (NGS) feasible for larger sample sizes. These new approaches have been lauded as unbiased but criticized as relative (rather than quantitative) measures of DNA methylation. The depth of genome coverage will be able to provide the large amount of information needed to detect subtle changes from multiple sources. Integration of these data with ongoing studies of biomarkers in other neurodegenerative diseases and in non- diseased aging populations will help elucidate the specific epigenetic changes associated with LOAD, providing a foundation for prevention and treatment of this disease.

20 FIGURES

Figure 1.1. Conceptual diagram describing the relationship between environmental exposures, including to the heavy metal lead, with the development of late-onset Alzheimer’s disease. There is a complex interplay of genetics and epigenetic programming. Epidemiologic cohort studies can be designed to study different stages in the life course leading to disease development.

21 REFERENCES

Almeida OP, Hulse GK, Lawrence D, Flicker L. 2002. Smoking as a risk factor for Alzheimer's disease: contrasting evidence from a systematic review of case-control and cohort studies. Addiction 97(1): 15-28. Anstey KJ, Cherbuin N, Budge M, Young J. 2011. Body mass index in midlife and late-life as a risk factor for dementia: a meta-analysis of prospective studies. Obes Rev 12(501): e426-437. Anway MD, Cupp AS, Uzumcu M, Skinner MK. 2005. Epigenetic transgenerational actions of endocrine disruptors and male fertility. Science 308(5727): 1466-1469. Ashford JW. 2004. APOE genotype effects on Alzheimer's disease onset and epidemiology. J Mol Neurosci 23(3): 157-165. Baldi I, Lebailly P, Mohammed-Brahim B, Letenneur L, Dartigues JF, Brochard P. 2003. Neurodegenerative diseases and exposure to pesticides in the elderly. Am J Epidemiol 157(5): 409-414. Barbosa F, Jr., Tanus-Santos JE, Gerlach RF, Parsons PJ. 2005. A critical review of biomarkers used for monitoring human exposure to lead: advantages, limitations, and future needs. Environ Health Perspect 113(12): 1669-1674. Barker DJ. 2004. The developmental origins of chronic adult disease. Acta Paediatr Suppl 93(446): 26-33. Barker DJ, Bull AR, Osmond C, Simmonds SJ. 1990. Fetal and placental size and risk of hypertension in adult life. BMJ 301(6746): 259-262. Barker DJ, Osmond C. 1986. Infant mortality, childhood nutrition, and ischaemic heart disease in England and Wales. Lancet 1(8489): 1077-1081. Barrachina M, Ferrer I. 2009. DNA methylation of Alzheimer disease and tauopathy-related genes in postmortem brain. J Neuropathol Exp Neurol 68(8): 880-891. Barry PSI, Mossman DB. 1970. Lead concentrations in human tissues. British Journal of Industrial Medicine 27: 339-351. Basha MR, Murali M, Siddiqi HK, Ghosal K, Siddiqi OK, Lashuel HA, Ge YW, Lahiri DK, Zawia NH. 2005a. Lead (Pb) exposure and its effect on APP proteolysis and Abeta aggregation. FASEB J 19(14): 2083-2084. Basha MR, Wei W, Bakheet SA, Benitez N, Siddiqi HK, Ge YW, Lahiri DK, Zawia NH. 2005b. The fetal basis of amyloidogenesis: exposure to lead and latent overexpression of amyloid precursor protein and beta-amyloid in the aging brain. J Neurosci 25(4): 823-829.

22 Belakavadi M, Dell J, Grover GJ, Fondell JD. 2011. Thyroid hormone suppression of beta-amyloid precursor protein gene expression in the brain involves multiple epigenetic regulatory events. Mol Cell Endocrinol 339(1-2): 72-80. Bergvall N, Iliadou A, Johansson S, de Faire U, Kramer MS, Pawitan Y, Pedersen NL, Lichtenstein P, Cnattingius S. 2007. Genetic and shared environmental factors do not confound the association between birth weight and hypertension: a study among Swedish twins. Circulation 115(23): 2931-2938. Bertram L. 2009. Alzheimer's disease genetics current status and future perspectives. Int Rev Neurobiol 84: 167-184. Bjornsson HT, Sigurdsson MI, Fallin MD, Irizarry RA, Aspelund T, Cui H, Yu W, Rongione MA, Ekstrom TJ, Harris TB, Launer LJ, Eiriksdottir G, Leppert MF, Sapienza C, Gudnason V, Feinberg AP. 2008. Intra-individual change over time in DNA methylation with familial clustering. JAMA 299(24): 2877-2883. Bolin CM, Basha R, Cox D, Zawia NH, Maloney B, Lahiri DK, Cardozo-Pelaez F. 2006. Exposure to lead and the developmental origin of oxidative DNA damage in the aging brain. FASEB J 20(6): 788-790. Borenstein AR, Copenhaver CI, Mortimer JA. 2006. Early-life risk factors for Alzheimer disease. Alzheimer Dis Assoc Disord 20(1): 63-72. Bottiglieri T, Godfrey P, Flynn T, Carney MW, Toone BK, Reynolds EH. 1990. Cerebrospinal fluid S-adenosylmethionine in depression and dementia: effects of treatment with parenteral and oral S-adenosylmethionine. J Neurol Neurosurg Psychiatry 53(12): 1096-1098. Brohede J, Rinde M, Winblad B, Graff C. 2010. A DNA methylation study of the amyloid precursor protein gene in several brain regions from patients with familial Alzheimer disease. J Neurogenet 24(4): 179-181. Brown DR. 2009. Brain proteins that mind metals: a neurodegenerative perspective. Dalton Trans(21): 4069-4076. Calderon-Garciduenas L, Reed W, Maronpot RR, Henriquez-Roldan C, Delgado- Chavez R, Calderon-Garciduenas A, Dragustinovis I, Franco-Lira M, Aragon-Flores M, Solt AC, Altenburg M, Torres-Jardon R, Swenberg JA. 2004. Brain inflammation and Alzheimer's-like pathology in individuals exposed to severe air pollution. Toxicol Pathol 32(6): 650-658. Calvanese V, Lara E, Kahn A, Fraga MF. 2009. The role of epigenetics in aging and age-related diseases. Ageing Res Rev 8(4): 268-276.

23 Cao XJ, Huang SH, Wang M, Chen JT, Ruan DY. 2008. S-adenosyl-L- methionine improves impaired hippocampal long-term potentiation and water maze performance induced by developmental lead exposure in rats. Eur J Pharmacol 595(1-3): 30-34. Cerda S, Weitzman SA. 1997. Influence of oxygen radical injury on DNA methylation. Mutat Res 386(2): 141-152. Chen T, Li YY, Zhang JL, Xu B, Lin Y, Wang CX, Guan WC, Wang YJ, Xu SQ. 2011. Protective effect of C(60) -methionine derivate on lead-exposed human SH-SY5Y neuroblastoma cells. J Appl Toxicol 31(3): 255-261. Cheng Y, Schwartz J, Sparrow D, Aro A, Weiss ST, Hu H. 2001. Bone lead and blood lead levels in relation to baseline blood pressure and the prospective development of hypertension: the Normative Aging Study. Am J Epidemiol 153(2): 164-171. Chettle DR. 2005. Three decades of in vivo x-ray fluorescence of lead in bone. X- Ray Spectrometry 34(5): 446-450. Chouliaras L, Rutten BP, Kenis G, Peerbooms O, Visser PJ, Verhey F, van Os J, Steinbusch HW, van den Hove DL. 2010. Epigenetic regulation in the pathophysiology of Alzheimer's disease. Prog Neurobiol 90(4): 498-510. Clarke R, Smith AD, Jobst KA, Refsum H, Sutton L, Ueland PM. 1998. Folate, vitamin B12, and serum total homocysteine levels in confirmed Alzheimer disease. Arch Neurol 55(11): 1449-1455. Coon S, Stark A, Peterson E, Gloi A, Kortsha G, Pounds J, Chettle D, Gorell J. 2006. Whole-body lifetime occupational lead exposure and risk of Parkinson's disease. Environ Health Perspect 114(12): 1872-1876. De Meyer G, Shapiro F, Vanderstichele H, Vanmechelen E, Engelborghs S, De Deyn PP, Coart E, Hansson O, Minthon L, Zetterberg H, Blennow K, Shaw L, Trojanowski JQ. 2010. Diagnosis-independent Alzheimer disease biomarker signature in cognitively normal elderly people. Arch Neurol 67(8): 949-956. DeBaun MR, Niemitz EL, McNeil DE, Brandenburg SA, Lee MP, Feinberg AP. 2002. Epigenetic alterations of H19 and LIT1 distinguish patients with Beckwith-Wiedemann syndrome with cancer and birth defects. Am J Hum Genet 70(3): 604-611. Dolinoy DC, Weidman JR, Jirtle RL. 2007. Epigenetic gene regulation: linking early developmental environment to adult disease. Reprod Toxicol 23(3): 297-307. Dubois B, Feldman HH, Jacova C, Dekosky ST, Barberger-Gateau P, Cummings J, Delacourte A, Galasko D, Gauthier S, Jicha G, Meguro K, O'Brien J,

24 Pasquier F, Robert P, Rossor M, Salloway S, Stern Y, Visser PJ, Scheltens P. 2007. Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 6(8): 734- 746. Edwards M, Triantafyllidou S, Best D. 2009. Elevated blood lead in young children due to lead-contaminated drinking water: Washington, DC, 2001- 2004. Environ Sci Technol 43(5): 1618-1623. Ertekin-Taner N. 2010. Genetics of Alzheimer disease in the pre- and post- GWAS era. Alzheimers Res Ther 2(1): 3. Esteller M. 2008. Epigenetics in cancer. N Engl J Med 358(11): 1148-1159. Faulk C, Dolinoy DC. 2011. Timing is everything: The when and how of environmentally induced changes in the epigenome of animals. Epigenetics 6(7). Feinberg AP, Irizarry RA. 2010. Evolution in health and medicine Sackler colloquium: Stochastic epigenetic variation as a driving force of development, evolutionary adaptation, and disease. Proc Natl Acad Sci U S A 107 Suppl 1: 1757-1764. Fewtrell LJ, Pruss-Ustun A, Landrigan P, Ayuso-Mateos JL. 2004. Estimating the global burden of disease of mild mental retardation and cardiovascular diseases from environmental lead exposure. Environ Res 94(2): 120-133. Fraga MF. 2009. Genetic and epigenetic regulation of aging. Curr Opin Immunol 21(4): 446-453. Fraga MF, Esteller M. 2007. Epigenetics and aging: the targets and the marks. Trends Genet 23(8): 413-418. Fratiglioni L, Wang HX. 2007. Brain reserve hypothesis in dementia. J Alzheimers Dis 12(1): 11-22. Frisardi V, Solfrizzi V, Capurso C, Kehoe PG, Imbimbo BP, Santamato A, Dellegrazie F, Seripa D, Pilotto A, Capurso A, Panza F. 2010. Aluminum in the diet and Alzheimer's disease: from current epidemiology to possible disease-modifying treatment. J Alzheimers Dis 20(1): 17-30. Fuso A, Seminara L, Cavallaro RA, D'Anselmi F, Scarpa S. 2005. S- adenosylmethionine/homocysteine cycle alterations modify DNA methylation status with consequent deregulation of PS1 and BACE and beta-amyloid production. Mol Cell Neurosci 28(1): 195-204. Garcia F, Ortega A, Domingo JL, Corbella J. 2001. Accumulation of metals in autopsy tissues of subjects living in Tarragona County, Spain. J Environ Sci Health A Tox Hazard Subst Environ Eng 36(9): 1767-1786.

25 Gatz M, Pedersen NL, Berg S, Johansson B, Johansson K, Mortimer JA, Posner SF, Viitanen M, Winblad B, Ahlbom A. 1997. Heritability for Alzheimer's disease: the study of dementia in Swedish twins. J Gerontol A Biol Sci Med Sci 52(2): M117-125. Gatz M, Reynolds CA, Fratiglioni L, Johansson B, Mortimer JA, Berg S, Fiske A, Pedersen NL. 2006. Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry 63(2): 168-174. Gerhardsson L, Lundh T, Minthon L, Londos E. 2008. Metal concentrations in plasma and cerebrospinal fluid in patients with Alzheimer's disease. Dement Geriatr Cogn Disord 25(6): 508-515. Grosse SD, Matte TD, Schwartz J, Jackson RJ. 2002. Economic gains resulting from the reduction in children's exposure to lead in the United States. Environ Health Perspect 110(6): 563-569. Guallar E, Silbergeld EK, Navas-Acien A, Malhotra S, Astor BC, Sharrett AR, Schwartz BS. 2006. Confounding of the relation between homocysteine and peripheral arterial disease by lead, cadmium, and renal function. Am J Epidemiol 163(8): 700-708. Hanson M, Godfrey KM, Lillycrop KA, Burdge GC, Gluckman PD. 2011. Developmental plasticity and developmental origins of non-communicable disease: Theoretical considerations and epigenetic mechanisms. Prog Biophys Mol Biol 106(1): 272-280. Hardy J. 1997. Amyloid, the presenilins and Alzheimer's disease. Trends Neurosci 20(4): 154-159. Harold D, Abraham R, Hollingworth P, Sims R, Gerrish A, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Williams A, Jones N, Thomas C, Stretton A, Morgan AR, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC, Gill M, Lawlor B, Lynch A, Morgan K, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Love S, Kehoe PG, Hardy J, Mead S, Fox N, Rossor M, Collinge J, Maier W, Jessen F, Schurmann B, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Hull M, Rujescu D, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Carrasquillo MM, Pankratz VS, Younkin SG, Holmans PA, O'Donovan M, Owen MJ, Williams J. 2009. Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer's disease. Nat Genet 41(10): 1088-1093.

26 Helmer C, Peuchant E, Letenneur L, Bourdel-Marchasson I, Larrieu S, Dartigues JF, Dubourg L, Thomas MJ, Barberger-Gateau P. 2003. Association between antioxidant nutritional indicators and the incidence of dementia: results from the PAQUID prospective cohort study. Eur J Clin Nutr 57(12): 1555-1561. Hernandez-Avila M, Smith D, Meneses F, Sanin LH, Hu H. 1998. The influence of bone and blood lead on plasma lead levels in environmentally exposed adults. Environ Health Perspect 106(8): 473-477. Hollingworth P, Harold D, Sims R, Gerrish A, Lambert JC, Carrasquillo MM, Abraham R, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Jones N, Stretton A, Thomas C, Richards A, Ivanov D, Widdowson C, Chapman J, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC, Gill M, Lawlor B, Lynch A, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Beaumont H, Warden D, Wilcock G, Love S, Kehoe PG, Hooper NM, Vardy ER, Hardy J, Mead S, Fox NC, Rossor M, Collinge J, Maier W, Jessen F, Ruther E, Schurmann B, Heun R, Kolsch H, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Gallacher J, Hull M, Rujescu D, Giegling I, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Pankratz VS, Sando SB, Aasly JO, Barcikowska M, Wszolek ZK, Dickson DW, Graff-Radford NR, Petersen RC, van Duijn CM, Breteler MM, Ikram MA, Destefano AL, Fitzpatrick AL, Lopez O, Launer LJ, Seshadri S, Berr C, Campion D, Epelbaum J, Dartigues JF, Tzourio C, Alperovitch A, Lathrop M, Feulner TM, Friedrich P, Riehle C, Krawczak M, Schreiber S, Mayhaus M, Nicolhaus S, Wagenpfeil S, Steinberg S, Stefansson H, Stefansson K, Snaedal J, Bjornsson S, Jonsson PV, Chouraki V, Genier-Boley B, Hiltunen M, Soininen H, Combarros O, Zelenika D, Delepine M, Bullido MJ, Pasquier F, Mateo I, Frank-Garcia A, Porcellini E, Hanon O, Coto E, Alvarez V, Bosco P, Siciliano G, Mancuso M, Panza F, Solfrizzi V, Nacmias B, Sorbi S, Bossu P, Piccardi P, Arosio B, Annoni G, Seripa D, Pilotto A, Scarpini E, Galimberti D, Brice A, Hannequin D, Licastro F, Jones L, Holmans PA, Jonsson T, Riemenschneider M, Morgan K, Younkin SG, Owen MJ, O'Donovan M, Amouyel P, Williams J. 2011. Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer's disease. Nat Genet. Horike S, Cai S, Miyano M, Cheng JF, Kohwi-Shigematsu T. 2005. Loss of silent- chromatin looping and impaired imprinting of DLX5 in Rett syndrome. Nat Genet 37(1): 31-40.

27 Hu H, Aro A, Payton M, Korrick S, Sparrow D, Weiss ST, Rotnitzky A. 1996. The relationship of bone and blood lead to hypertension. The Normative Aging Study. JAMA 275(15): 1171-1176. Hu H, Aro A, Rotnitzky A. 1995. Bone lead measured by X-ray fluorescence: epidemiologic methods. Environ Health Perspect 103 Suppl 1: 105-110. Hu H, Milder FL, Burger DE. 1989. X-ray fluorescence: issues surrounding the application of a new tool for measuring burden of lead. Environ Res 49(2): 295-317. Hu H, Rabinowitz M, Smith D. 1998. Bone lead as a biological marker in epidemiologic studies of chronic toxicity: conceptual paradigms. Environ Health Perspect 106(1): 1-8. Hu H, Shih R, Rothenberg S, Schwartz BS. 2007. The epidemiology of lead toxicity in adults: measuring dose and consideration of other methodologic issues. Environ Health Perspect 115(3): 455-462. Hu H, Tosteson T, Aufderheide AC, Wittmers L, Burger DE, Milder FL, Schidlovsky G, Jones KW. 1990. Distribution of lead in human bone: I. Atomic absorption measurements. Basic Life Sci 55: 267-274. Huang T, Carlson M, Fitzpatrick A, Kuller L, Fried L, Zandi P. 2008. Knee height and arm span: a reflection of early life environment and risk of dementia. Neurology 70(19, Part 2): 1818. Jablonka E. 2004. Epigenetic epidemiology. Int J Epidemiol 33(5): 929-935. Jack CR, Jr., Petersen RC, O'Brien PC, Tangalos EG. 1992. MR-based hippocampal volumetry in the diagnosis of Alzheimer's disease. Neurology 42(1): 183-188. Jain NB, Potula V, Schwartz J, Vokonas PS, Sparrow D, Wright RO, Nie H, Hu H. 2007. Lead levels and ischemic heart disease in a prospective study of middle-aged and elderly men: the VA Normative Aging Study. Environ Health Perspect 115(6): 871-875. Jeong SK, Kim JM, Kweon SS, Shin MH, Seo MW, Kim YH. 2005. Does arm length indicate cognitive and functional reserve? International journal of geriatric psychiatry 20(5): 406-412. Jones PA, Taylor SM. 1980. Cellular differentiation, cytidine analogs and DNA methylation. Cell 20(1): 85-93. Jung CH, Matsuto T, Tanaka N, Okada T. 2004. Metal distribution in incineration residues of municipal solid waste (MSW) in Japan. Waste Management 24(4): 381-391.

28 Kamel F, Umbach DM, Hu H, Munsat TL, Shefner JM, Taylor JA, Sandler DP. 2005. Lead exposure as a risk factor for amyotrophic lateral sclerosis. Neurodegener Dis 2(3-4): 195-201. Kamel F, Umbach DM, Lehman TA, Park LP, Munsat TL, Shefner JM, Sandler DP, Hu H, Taylor JA. 2003. Amyotrophic lateral sclerosis, lead, and genetic susceptibility: polymorphisms in the delta-aminolevulinic acid dehydratase and vitamin D receptor genes. Environ Health Perspect 111(10): 1335-1339. Kamel F, Umbach DM, Munsat TL, Shefner JM, Hu H, Sandler DP. 2002. Lead exposure and amyotrophic lateral sclerosis. Epidemiology 13(3): 311-319. Kaul B, Sandhu RS, Depratt C, Reyes F. 1999. Follow-up screening of lead- poisoned children near an auto battery recycling plant, Haina, Dominican Republic. Environ Health Perspect 107(11): 917-920. Ke Q, Davidson T, Chen H, Kluz T, Costa M. 2006. Alterations of histone modifications and transgene silencing by nickel chloride. Carcinogenesis 27(7): 1481-1488. Kim J, Basak JM, Holtzman DM. 2009. The role of apolipoprotein E in Alzheimer's disease. Neuron 63(3): 287-303. Klunk WE, Engler H, Nordberg A, Wang Y, Blomqvist G, Holt DP, Bergstrom M, Savitcheva I, Huang GF, Estrada S, Ausen B, Debnath ML, Barletta J, Price JC, Sandell J, Lopresti BJ, Wall A, Koivisto P, Antoni G, Mathis CA, Langstrom B. 2004. Imaging brain amyloid in Alzheimer's disease with Pittsburgh Compound-B. Ann Neurol 55(3): 306-319. Korrick SA, Hunter DJ, Rotnitzky A, Hu H, Speizer FE. 1999. Lead and hypertension in a sample of middle-aged women. Am J Public Health 89(3): 330-335. Kovatsi L, Georgiou E, Ioannou A, Haitoglou C, Tzimagiorgis G, Tsoukali H, Kouidou S. 2010. p16 promoter methylation in Pb2+ -exposed individuals. Clin Toxicol (Phila) 48(2): 124-128. Kukull WA, Larson EB, Bowen JD, McCormick WC, Teri L, Pfanschmidt ML, Thompson JD, O'Meara ES, Brenner DE, van Belle G. 1995. Solvent exposure as a risk factor for Alzheimer's disease: a case-control study. Am J Epidemiol 141(11): 1059-1071; discussion 1072-1059. Lambert JC, Heath S, Even G, Campion D, Sleegers K, Hiltunen M, Combarros O, Zelenika D, Bullido MJ, Tavernier B, Letenneur L, Bettens K, Berr C, Pasquier F, Fievet N, Barberger-Gateau P, Engelborghs S, De Deyn P, Mateo I, Franck A, Helisalmi S, Porcellini E, Hanon O, de Pancorbo MM, Lendon C, Dufouil C, Jaillard C, Leveillard T, Alvarez V, Bosco P, Mancuso M, Panza F, Nacmias B, Bossu P, Piccardi P, Annoni G, Seripa

29 D, Galimberti D, Hannequin D, Licastro F, Soininen H, Ritchie K, Blanche H, Dartigues JF, Tzourio C, Gut I, Van Broeckhoven C, Alperovitch A, Lathrop M, Amouyel P. 2009. Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer's disease. Nat Genet 41(10): 1094-1099. Landrigan PJ, Sonawane B, Butler RN, Trasande L, Callan R, Droller D. 2005. Early environmental origins of neurodegenerative disease in later life. Environ Health Perspect 113(9): 1230-1233. Lanphear BP, Byrd RS, Auinger P, Schaffer SJ. 1998. Community characteristics associated with elevated blood lead levels in children. Pediatrics 101(2): 264-271. Larrieu S, Letenneur L, Helmer C, Dartigues JF, Barberger-Gateau P. 2004. Nutritional factors and risk of incident dementia in the PAQUID longitudinal cohort. J Nutr Health Aging 8(3): 150-154. Laurin D, Verreault R, Lindsay J, MacPherson K, Rockwood K. 2001. Physical activity and risk of cognitive impairment and dementia in elderly persons. Arch Neurol 58(3): 498-504. Lester-Coll N, Rivera EJ, Soscia SJ, Doiron K, Wands JR, de la Monte SM. 2006. Intracerebral streptozotocin model of type 3 diabetes: relevance to sporadic Alzheimer's disease. J Alzheimers Dis 9(1): 13-33. Lewin MD, Sarasua S, Jones PA. 1999. A multivariate linear regression model for predicting children's blood lead levels based on soil lead levels: A study at four superfund sites. Environ Res 81(1): 52-61. Li G, Shofer JB, Rhew IC, Kukull WA, Peskind ER, McCormick W, Bowen JD, Schellenberg GD, Crane PK, Breitner JC, Larson EB. 2010. Age-varying association between statin use and incident Alzheimer's disease. J Am Geriatr Soc 58(7): 1311-1317. Li J, Wang YJ, Zhang M, Xu ZQ, Gao CY, Fang CQ, Yan JC, Zhou HD. 2011. Vascular risk factors promote conversion from mild cognitive impairment to Alzheimer disease. Neurology. Li YJ, Scott WK, Hedges DJ, Zhang F, Gaskell PC, Nance MA, Watts RL, Hubble JP, Koller WC, Pahwa R, Stern MB, Hiner BC, Jankovic J, Allen FA, Jr., Goetz CG, Mastaglia F, Stajich JM, Gibson RA, Middleton LT, Saunders AM, Scott BL, Small GW, Nicodemus KK, Reed AD, Schmechel DE, Welsh-Bohmer KA, Conneally PM, Roses AD, Gilbert JR, Vance JM, Haines JL, Pericak-Vance MA. 2002. Age at onset in two common neurodegenerative diseases is genetically controlled. Am J Hum Genet 70(4): 985-993.

30 Li YY, Chen T, Wan Y, Xu SQ. 2010. Lead exposure in pheochromocytoma cells induces persistent changes in amyloid precursor protein gene methylation patterns. Environ Toxicol. Lindsay J, Laurin D, Verreault R, Hebert R, Helliwell B, Hill GB, McDowell I. 2002. Risk factors for Alzheimer's disease: a prospective analysis from the Canadian Study of Health and Aging. Am J Epidemiol 156(5): 445-453. Luchsinger JA, Tang MX, Stern Y, Shea S, Mayeux R. 2001. Diabetes mellitus and risk of Alzheimer's disease and dementia with stroke in a multiethnic cohort. American Journal of Epidemiology 154(7): 635. Mandel S, Amit T, Bar-Am O, Youdim MB. 2007. Iron dysregulation in Alzheimer's disease: multimodal brain permeable iron chelating drugs, possessing neuroprotective-neurorescue and amyloid precursor protein- processing regulatory activities as therapeutic agents. Prog Neurobiol 82(6): 348-360. Marsit CJ, Karagas MR, Danaee H, Liu M, Andrew A, Schned A, Nelson HH, Kelsey KT. 2006. Carcinogen exposure and gene promoter hypermethylation in bladder cancer. Carcinogenesis 27(1): 112-116. Mastroeni D, Grover A, Delvaux E, Whiteside C, Coleman PD, Rogers J. 2008. Epigenetic changes in Alzheimer's disease: Decrements in DNA methylation. Neurobiol Aging. Mastroeni D, Grover A, Delvaux E, Whiteside C, Coleman PD, Rogers J. 2011. Epigenetic mechanisms in Alzheimer's disease. Neurobiol Aging. Mastroeni D, McKee A, Grover A, Rogers J, Coleman PD. 2009. Epigenetic differences in cortical neurons from a pair of monozygotic twins discordant for Alzheimer's disease. PLoS One 4(8): e6617. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. 1984. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology 34(7): 939-944. Meyer PA, Brown MJ, Falk H. 2008. Global approach to reducing lead exposure and poisoning. Mutat Res 659(1-2): 166-175. Mill J. 2011. Toward an integrated genetic and epigenetic approach to Alzheimer's disease. Neurobiol Aging. Miller DB, O'Callaghan JP. 2008. Do early-life insults contribute to the late-life development of Parkinson and Alzheimer diseases? Metabolism 57 Suppl 2: S44-49.

31 Minoshima S, Frey KA, Koeppe RA, Foster NL, Kuhl DE. 1995. A diagnostic approach in Alzheimer's disease using three-dimensional stereotactic surface projections of fluorine-18-FDG PET. J Nucl Med 36(7): 1238-1248. Moceri VM, Kukull WA, Emanuel I, van Belle G, Larson EB. 2000. Early-life risk factors and the development of Alzheimer's disease. Neurology 54(2): 415-420. Morris KV. 2011. The emerging role of RNA in the regulation of gene transcription in human cells. Semin Cell Dev Biol. Morris MC. 2009. The role of nutrition in Alzheimer's disease: epidemiological evidence. Eur J Neurol 16 Suppl 1: 1-7. Naj AC, Jun G, Beecham GW, Wang LS, Vardarajan BN, Buros J, Gallins PJ, Buxbaum JD, Jarvik GP, Crane PK, Larson EB, Bird TD, Boeve BF, Graff- Radford NR, De Jager PL, Evans D, Schneider JA, Carrasquillo MM, Ertekin-Taner N, Younkin SG, Cruchaga C, Kauwe JS, Nowotny P, Kramer P, Hardy J, Huentelman MJ, Myers AJ, Barmada MM, Demirci FY, Baldwin CT, Green RC, Rogaeva E, George-Hyslop PS, Arnold SE, Barber R, Beach T, Bigio EH, Bowen JD, Boxer A, Burke JR, Cairns NJ, Carlson CS, Carney RM, Carroll SL, Chui HC, Clark DG, Corneveaux J, Cotman CW, Cummings JL, Decarli C, Dekosky ST, Diaz-Arrastia R, Dick M, Dickson DW, Ellis WG, Faber KM, Fallon KB, Farlow MR, Ferris S, Frosch MP, Galasko DR, Ganguli M, Gearing M, Geschwind DH, Ghetti B, Gilbert JR, Gilman S, Giordani B, Glass JD, Growdon JH, Hamilton RL, Harrell LE, Head E, Honig LS, Hulette CM, Hyman BT, Jicha GA, Jin LW, Johnson N, Karlawish J, Karydas A, Kaye JA, Kim R, Koo EH, Kowall NW, Lah JJ, Levey AI, Lieberman AP, Lopez OL, Mack WJ, Marson DC, Martiniuk F, Mash DC, Masliah E, McCormick WC, McCurry SM, McDavid AN, McKee AC, Mesulam M, Miller BL, Miller CA, Miller JW, Parisi JE, Perl DP, Peskind E, Petersen RC, Poon WW, Quinn JF, Rajbhandary RA, Raskind M, Reisberg B, Ringman JM, Roberson ED, Rosenberg RN, Sano M, Schneider LS, Seeley W, Shelanski ML, Slifer MA, Smith CD, Sonnen JA, Spina S, Stern RA, Tanzi RE, Trojanowski JQ, Troncoso JC, Van Deerlin VM, Vinters HV, Vonsattel JP, Weintraub S, Welsh-Bohmer KA, Williamson J, Woltjer RL, Cantwell LB, Dombroski BA, Beekly D, Lunetta KL, Martin ER, Kamboh MI, Saykin AJ, Reiman EM, Bennett DA, Morris JC, Montine TJ, Goate AM, Blacker D, Tsuang DW, Hakonarson H, Kukull WA, Foroud TM, Haines JL, Mayeux R, Pericak-Vance MA, Farrer LA, Schellenberg GD. 2011. Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer's disease. Nat Genet. Nedwed T, Clifford DA. 1997. A survey of lead battery recycling sites and soil remediation processes. Waste Management 17(4): 257-269.

32 Nee LE, Lippa CF. 1999. Alzheimer's disease in 22 twin pairs--13-year follow-up: hormonal, infectious and traumatic factors. Dement Geriatr Cogn Disord 10(2): 148-151. Niculescu MD, Craciunescu CN, Zeisel SH. 2005. Gene expression profiling of choline-deprived neural precursor cells isolated from mouse brain. Brain Res Mol Brain Res 134(2): 309-322. Niculescu MD, Craciunescu CN, Zeisel SH. 2006. Dietary choline deficiency alters global and gene-specific DNA methylation in the developing hippocampus of mouse fetal brains. FASEB J 20(1): 43-49. Nunez-Iglesias J, Liu CC, Morgan TE, Finch CE, Zhou XJ. 2010. Joint genome- wide profiling of miRNA and mRNA expression in Alzheimer's disease cortex reveals altered miRNA regulation. PLoS One 5(2): e8898. Ogawa O, Zhu X, Lee HG, Raina A, Obrenovich ME, Bowser R, Ghanbari HA, Castellani RJ, Perry G, Smith MA. 2003. Ectopic localization of phosphorylated histone H3 in Alzheimer's disease: a mitotic catastrophe? Acta neuropathologica 105(5): 524-528. Park SK, Mukherjee B, Xia X, Sparrow D, Weisskopf MG, Nie H, Hu H. 2009. Bone lead level prediction models and their application to examine the relationship of lead exposure and hypertension in the Third National Health and Nutrition Examination Survey. J Occup Environ Med 51(12): 1422-1436. Payton M, Riggs KM, Spiro A, 3rd, Weiss ST, Hu H. 1998. Relations of bone and blood lead to cognitive function: the VA Normative Aging Study. Neurotoxicol Teratol 20(1): 19-27. Pilsner JR, Hu H, Ettinger A, Sanchez BN, Wright RO, Cantonwine D, Lazarus A, Lamadrid-Figueroa H, Mercado-Garcia A, Tellez-Rojo MM, Hernandez- Avila M. 2009. Influence of prenatal lead exposure on genomic methylation of cord blood DNA. Environ Health Perspect 117(9): 1466- 1471. Pirkle JL, Brody DJ, Gunter EW, Kramer RA, Paschal DC, Flegal KM, Matte TD. 1994. The decline in blood lead levels in the United States. The National Health and Nutrition Examination Surveys (NHANES). JAMA 272(4): 284- 291. Plassman BL, Havlik RJ, Steffens DC, Helms MJ, Newman TN, Drosdick D, Phillips C, Gau BA, Welsh-Bohmer KA, Burke JR, Guralnik JM, Breitner JC. 2000. Documented head injury in early adulthood and risk of Alzheimer's disease and other dementias. Neurology 55(8): 1158-1166. Pounds JG, Long GJ, Rosen JF. 1991. Cellular and molecular toxicity of lead in bone. Environ Health Perspect 91: 17-32.

33 Rabinowitz MB. 1991. Toxicokinetics of bone lead. Environ Health Perspect 91: 33-37. Rabinowitz MB, Wetherill GW, Kopple JD. 1976. Kinetic analysis of lead metabolism in healthy humans. J Clin Invest 58(2): 260-270. Rabito FA, Shorter C, White LE. 2003. Lead levels among children who live in public housing. Epidemiology 14(3): 263-268. Ravelli AC, van der Meulen JH, Michels RP, Osmond C, Barker DJ, Hales CN, Bleker OP. 1998. Glucose tolerance in adults after prenatal exposure to famine. Lancet 351(9097): 173-177. Ray S, Britschgi M, Herbert C, Takeda-Uchimura Y, Boxer A, Blennow K, Friedman LF, Galasko DR, Jutel M, Karydas A, Kaye JA, Leszek J, Miller BL, Minthon L, Quinn JF, Rabinovici GD, Robinson WH, Sabbagh MN, So YT, Sparks DL, Tabaton M, Tinklenberg J, Yesavage JA, Tibshirani R, Wyss-Coray T. 2007. Classification and prediction of clinical Alzheimer's diagnosis based on plasma signaling proteins. Nat Med 13(11): 1359- 1362. Rezai-Zadeh K, Shytle D, Sun N, Mori T, Hou H, Jeanniton D, Ehrhart J, Townsend K, Zeng J, Morgan D. 2005. Green tea epigallocatechin-3- gallate (EGCG) modulates amyloid precursor protein cleavage and reduces cerebral amyloidosis in Alzheimer transgenic mice. The Journal of neuroscience 25(38): 8807. Reznichenko L, Amit T, Zheng H, Avramovich Tirosh Y, Youdim M, Mandel S. 2006. Reduction of iron regulated amyloid precursor protein and amyloid peptide by (–) epigallocatechin 3 gallate in cell cultures: implications for iron chelation in Alzheimer's disease. Journal of neurochemistry 97(2): 527-536. Ross MG, Desai M, Khorram O, McKnight RA, Lane RH, Torday J. 2007. Gestational programming of offspring obesity: a potential contributor to Alzheimer's disease. Curr Alzheimer Res 4(2): 213-217. Rovio S, Kareholt I, Helkala EL, Viitanen M, Winblad B, Tuomilehto J, Soininen H, Nissinen A, Kivipelto M. 2005. Leisure-time physical activity at midlife and the risk of dementia and Alzheimer's disease. The Lancet Neurology 4(11): 705-711. Rusanen M, Kivipelto M, Quesenberry CP, Jr., Zhou J, Whitmer RA. 2011. Heavy smoking in midlife and long-term risk of Alzheimer disease and vascular dementia. Arch Intern Med 171(4): 333-339. Saczynski JS, Beiser A, Seshadri S, Auerbach S, Wolf PA, Au R. 2010. Depressive symptoms and risk of dementia: the Framingham Heart Study. Neurology 75(1): 35-41.

34 Salnikow K, Zhitkovich A. 2008. Genetic and epigenetic mechanisms in metal carcinogenesis and cocarcinogenesis: nickel, arsenic, and chromium. Chem Res Toxicol 21(1): 28-44. Sananbenesi F, Fischer A. 2009. The epigenetic bottleneck of neurodegenerative and psychiatric diseases. Biol Chem 390(11): 1145-1153. Santibanez M, Bolumar F, Garcia AM. 2007. Occupational risk factors in Alzheimer's disease: a review assessing the quality of published epidemiological studies. Occup Environ Med 64(11): 723-732. Savva GM, Stephan BC. 2010. Epidemiological studies of the effect of stroke on incident dementia: a systematic review. Stroke 41(1): e41-46. Schafer JH, Glass TA, Bressler J, Todd AC, Schwartz BS. 2005. Blood lead is a predictor of homocysteine levels in a population-based study of older adults. Environ Health Perspect 113(1): 31-35. Schwartz BS, Stewart WF, Bolla KI, Simon PD, Bandeen-Roche K, Gordon PB, Links JM, Todd AC. 2000. Past adult lead exposure is associated with longitudinal decline in cognitive function. Neurology 55(8): 1144-1150. Seshadri S, Beiser A, Selhub J, Jacques PF, Rosenberg IH, D'Agostino RB, Wilson PW, Wolf PA. 2002. Plasma homocysteine as a risk factor for dementia and Alzheimer's disease. N Engl J Med 346(7): 476-483. Sharp SI, Aarsland D, Day S, Sonnesyn H, Ballard C. 2011. Hypertension is a potential risk factor for vascular dementia: systematic review. Int J Geriatr Psychiatry 26(7): 661-669. Shcherbatykh I, Carpenter DO. 2007. The role of metals in the etiology of Alzheimer's disease. J Alzheimers Dis 11(2): 191-205. Shih RA, Hu H, Weisskopf MG, Schwartz BS. 2007. Cumulative lead dose and cognitive function in adults: a review of studies that measured both blood lead and bone lead. Environ Health Perspect 115(3): 483-492. Siegmund KD, Connor CM, Campan M, Long TI, Weisenberger DJ, Biniszkiewicz D, Jaenisch R, Laird PW, Akbarian S. 2007. DNA methylation in the human cerebral cortex is dynamically regulated throughout the life span and involves differentiated neurons. PLoS One 2(9): e895. Silbergeld EK. 1991. Lead in bone: implications for toxicology during pregnancy and lactation. Environ Health Perspect 91: 63-70. Silbergeld EK, Schwartz J, Mahaffey K. 1988. Lead and osteoporosis: mobilization of lead from bone in postmenopausal women. Environ Res 47(1): 79-94.

35 Sobel E, Dunn M, Davanipour Z, Qian Z, Chui HC. 1996. Elevated risk of Alzheimer's disease among workers with likely electromagnetic field exposure. Neurology 47(6): 1477-1481. Stern C, Munn Z. 2010. Cognitive leisure activities and their role in preventing dementia: a systematic review. Int J Evid Based Healthc 8(1): 2-17. Sun H, Zhou X, Chen H, Li Q, Costa M. 2009. Modulation of histone methylation and MLH1 gene silencing by hexavalent chromium. Toxicol Appl Pharmacol 237(3): 258-266. Takiguchi M, Achanzar WE, Qu W, Li G, Waalkes MP. 2003. Effects of cadmium on DNA-(Cytosine-5) methyltransferase activity and DNA methylation status during cadmium-induced cellular transformation. Exp Cell Res 286(2): 355-365. Terry MB, Delgado-Cruzata L, Vin-Raviv N, Wu HC, Santella RM. 2011. DNA methylation in white blood cells: Association with risk factors in epidemiologic studies. Epigenetics 6(7). Tohgi H, Utsugisawa K, Nagane Y, Yoshimura M, Genda Y, Ukitsu M. 1999a. Reduction with age in methylcytosine in the promoter region -224 approximately -101 of the amyloid precursor protein gene in autopsy human cortex. Brain Res Mol Brain Res 70(2): 288-292. Tohgi H, Utsugisawa K, Nagane Y, Yoshimura M, Ukitsu M, Genda Y. 1999b. The methylation status of cytosines in a tau gene promoter region alters with age to downregulate transcriptional activity in human cerebral cortex. Neurosci Lett 275(2): 89-92. Tong S, von Schirnding YE, Prapamontol T. 2000. Environmental lead exposure: a public health problem of global dimensions. Bull World Health Organ 78(9): 1068-1077. Toscano CD, Guilarte TR. 2005. Lead neurotoxicity: from exposure to molecular effects. Brain Res Brain Res Rev 49(3): 529-554. Urdinguio RG, Sanchez-Mut JV, Esteller M. 2009. Epigenetic mechanisms in neurological diseases: genes, syndromes, and therapies. Lancet Neurol 8(11): 1056-1072. Verreault R, Laurin D, Lindsay J, De Serres G. 2001. Past exposure to vaccines and subsequent risk of Alzheimer's disease. CMAJ 165(11): 1495-1498. von Lindern I, Spalinger S, Petroysan V, von Braun M. 2003. Assessing remedial effectiveness through the blood lead: soil/dust relationship at the Bunker Hill Superfund Site in the Silver Valley of Idaho. The Science of The Total Environment 303(1-2): 139-170.

36 Wadhwa PD, Buss C, Entringer S, Swanson JM. 2009. Developmental origins of health and disease: brief history of the approach and current focus on epigenetic mechanisms. Semin Reprod Med 27(5): 358-368. Wang FT, Hu H, Schwartz J, Weuve J, Spiro AS, Sparrow D, Nie H, Silverman EK, Weiss ST, Wright RO. 2007. Modifying effects of the HFE polymorphisms on the association between lead burden and cognitive decline. Environ Health Perspect 115(8): 1210-1215. Wang SC, Oelze B, Schumacher A. 2008. Age-specific epigenetic drift in late- onset Alzheimer's disease. PLoS One 3(7): e2698. Waterland RA, Garza C. 1999. Potential mechanisms of metabolic imprinting that lead to chronic disease. Am J Clin Nutr 69(2): 179-197. Weisskopf MG, Hu H, Sparrow D, Lenkinski RE, Wright RO. 2007. Proton magnetic resonance spectroscopic evidence of glial effects of cumulative lead exposure in the adult human hippocampus. Environ Health Perspect 115(4): 519-523. Weisskopf MG, Jain N, Nie H, Sparrow D, Vokonas P, Schwartz J, Hu H. 2009. A prospective study of bone lead concentration and death from all causes, cardiovascular diseases, and cancer in the Department of Veterans Affairs Normative Aging Study. Circulation 120(12): 1056-1064. Weisskopf MG, Weuve J, Nie H, Saint-Hilaire MH, Sudarsky L, Simon DK, Hersh B, Schwartz J, Wright RO, Hu H. 2010. Association of Cumulative Lead Exposure with Parkinson's Disease. Environ Health Perspect. Weisskopf MG, Wright RO, Schwartz J, Spiro A, 3rd, Sparrow D, Aro A, Hu H. 2004. Cumulative lead exposure and prospective change in cognition among elderly men: the VA Normative Aging Study. Am J Epidemiol 160(12): 1184-1193. Weuve J, Kelsey KT, Schwartz J, Bellinger D, Wright RO, Rajan P, Spiro A, Sparrow D, Aro A, Hu H. 2006. Delta-aminolevulinic acid dehydratase polymorphism and the relation between low level lead exposure and the Mini-Mental Status Examination in older men: the Normative Aging Study. Occupational and environmental medicine 63(11): 746. Whitmer RA, Gunderson EP, Barrett-Connor E, Quesenberry CP, Yaffe K. 2005. Obesity in middle age and future risk of dementia: a 27 year longitudinal population based study. BMJ 330(7504): 1360. Wittmers LE, Jr., Aufderheide AC, Wallgren J, Rapp G, Jr., Alich A. 1988. Lead in bone. IV. Distribution of lead in the human skeleton. Arch Environ Health 43(6): 381-391.

37 Wright RO, Schwartz J, Wright RJ, Bollati V, Tarantini L, Park SK, Hu H, Sparrow D, Vokonas P, Baccarelli A. 2010. Biomarkers of Lead Exposure and DNA Methylation within Retrotransposons. Environ Health Perspect. Wright RO, Tsaih SW, Schwartz J, Spiro A, 3rd, McDonald K, Weiss ST, Hu H. 2003. Lead exposure biomarkers and mini-mental status exam scores in older men. Epidemiology 14(6): 713-718. Wu J, Basha MR, Brock B, Cox DP, Cardozo-Pelaez F, McPherson CA, Harry J, Rice DC, Maloney B, Chen D, Lahiri DK, Zawia NH. 2008a. Alzheimer's disease (AD)-like pathology in aged monkeys after infantile exposure to environmental metal lead (Pb): evidence for a developmental origin and environmental link for AD. J Neurosci 28(1): 3-9. Wu J, Basha MR, Zawia NH. 2008b. The environment, epigenetics and amyloidogenesis. J Mol Neurosci 34(1): 1-7. Wu SC, Zhang Y. 2010. Active DNA demethylation: many roads lead to Rome. Nat Rev Mol Cell Biol 11(9): 607-620. Yang F, Lim GP, Begum AN, Ubeda OJ, Simmons MR, Ambegaokar SS, Chen PP, Kayed R, Glabe CG, Frautschy SA, Cole GM. 2005. Curcumin inhibits formation of amyloid beta oligomers and fibrils, binds plaques, and reduces amyloid in vivo. J Biol Chem 280(7): 5892-5901.

38 CHAPTER II Research Chapter 1

Lead exposure, B-vitamins, and plasma homocysteine in the VA Normative Aging Study

ABSTRACT

Background: Epidemiologic studies suggest that elevated circulating level of homocysteine (Hcy), a one-carbon metabolite, is a risk factor for cardiovascular and neurodegenerative diseases. Although a few cross-sectional studies have evaluated the influence of environmental toxicant exposures on Hcy levels, longitudinal studies and studies of the interplay of environmental and dietary factors are lacking.

Objectives: We examined the association of recent and cumulative exposure to lead with Hcy levels cross-sectionally and longitudinally. We also determined whether lead exposure’s association with Hcy varied by dietary intake of nutrients involved in one-carbon metabolism (folate and vitamins B6 and B12).

Methods: We followed 1,056 of the Normative Aging Study men (age 50-97 at visit 1) over 4 study visits (2,301 total observations) at which concurrent measures of blood lead and Hcy were collected. We determined baseline cumulative dose of lead via Cd109 K-shell X-ray fluorescence of the tibia and patella bones. We estimated cross-sectional differences in Hcy across levels of Pb exposure measures using generalized linear models. We also used mixed effects models to estimate differences in rate of change in Hcy over time associated with Pb exposure.

39 Results: Higher exposure to lead was associated with higher Hcy levels. An interquartile range (IQR) increment in tibia bone lead concentration (14µg/g) was associated with 3.64% higher Hcy at visit 1 (95% CI= 1.37-5.97). Similar results were found at other visits, but tibia Pb was not associated with trajectory in Hcy over time. An IQR increment in blood lead concentration (3 µg/dl) was associated with 8.14% higher Hcy (95% CI: 6.16-10.16) at visit 1, similar to the results at other visits. To put these findings in context, in our data, a 5-year increment in age corresponded to a 3.14% increase in Hcy (95% CI=2.88-7.87). The association between blood lead and Hcy was significantly larger among participants with lower dietary intakes of vitamins B6, B12, and folate.

Conclusion: Increasing levels of lead exposure were associated with elevated Hcy, and this relationship was stronger in individuals with low dietary folate, B6, and B12. Increased intake of folate, B6, and B12 may be an effective intervention on lead’s effects on Hcy.

KEYWORDS: aging, folate, homocysteine, lead exposure

40 INTRODUCTION

The post-war baby boom and lengthening life spans are fueling an unprecedented growth in the population of older adults worldwide, with widespread implications for stress on public health and medical infrastructures. Thus, increasing numbers of older adults are at risk of chronic diseases that require care and monitoring over decades. Several chronic diseases, including cardiovascular disease (CVD) and neurodegenerative diseases, share a common risk factor, elevated homocysteine (Hcy), which is measured in the blood or cerebrospinal fluid. Relatively little is known about the causes of elevated Hcy; in particular, the potential influence of environmental toxicant exposures and their interplay with dietary factors. Research on the regulation of Hcy may result in opportunities for intervention prior to onset of chronic disease.

Homocysteine (Hcy)

Hcy is a thiol-containing amino acid that is highly reactive and thus short- lived in the body (Jocelyn 1972). It is an intermediate in the one-carbon metabolism cycle, formed in the production of methionine, an important methyl donor for epigenetic modifications to nucleic acids and proteins (FIGURE 2.1). Though physiologically normal cellular processes produce and require Hcy at low levels, elevated Hcy is associated with toxicity. Accessible cysteinyl residues in cellular proteins can react with free Hcy, forming Hcy-protein thiol-thiol interactions, altering native protein conformation and function. In addition, free Hcy can cleave accessible disulfide bridges, damaging native protein confirmations (Krumdieck and Prince 2000). Biochemical damage is based on the duration and concentration of exposure to Hcy. Long-lived proteins can accumulate irreversible Hcy-related damages, making these mechanisms especially relevant to the chronic diseases and morbidity of aging.

Elevated Hcy is a risk factor for both CVD and neurodegeneration. In epidemiologic studies, moderately elevated Hcy is associated with CVD. In a meta-analysis of 30 retrospective or prospective studies, reduced plasma Hcy

41 was protective against ischemic heart disease (IHD) (OR=0.89 for 25% lower Hcy), and stroke (OR=0.81) (Collaboration 2002). Similarly, a 5-μmol/L increase in Hcy corresponded to elevated odds of IHD (OR=1.23) and stroke (OR=1.42) in a meta-analysis of 20 prospective studies (Wald et al. 2002). The suggested mechanisms linking Hcy and cardiovascular outcomes include impaired endothelial elasticity and the production of reactive oxygen species (Perla-Kajan et al. 2007).

Elevated Hcy is also associated with decline in multiple cognitive domains. In the Veteran’s Affairs Normative Aging Study (NAS) of older men, higher baseline Hcy was associated with reduced spatial copying and verbal recall over a three-year follow-up period (Tucker et al. 2005). Data from the Oxford Health Aging Project show high Hcy is associated with declines in general cognitive function as measured by the Mini-Mental Status Exam (MMSE) over a ten-year period (Clarke et al. 2007). In a two-year follow-up study in Korea, incidence of dementia increased across ascending quintile of Hcy at follow-up (Kim et al. 2008). In the Sacramento Area Latino Study on Aging, higher baseline Hcy was associated with increased risk of both incident dementia and cognitive impairment in the absence of dementia over the subsequent 4.5 years of follow- up (Haan et al. 2007). Hcy is also an important risk factor for Alzheimer’s disease (Seshadri et al. 2002).

Circulating Hcy can be lowered with dietary interventions, primarily therapy with folic acid, vitamin B12, and vitamin B6 (Appel et al. 2000). Several randomized control trials have evaluated the effects of these therapies on cardio- and cerebrovascular events, their results have been mixed. A meta-analysis of seventeen trials of individuals with preexisting cardiovascular or renal disease showed no differences between the Hcy-lowering treatment group and control group with respect to coronary heart disease, stroke, cardiovascular events, or all-cause mortality (Mei et al. 2010). Dietary changes can reduce Hcy, but in individuals with pre-existing conditions, hypothesized improvements in health outcome do not seem to occur with the reduction (Mei et al. 2010).

42 Lead Exposure

The heavy metal lead (Pb), is a well-established and ubiquitous toxicant. The molecular mechanisms of lead’s toxicity in the human body are numerous but incompletely characterized. Pb binds free sulfhydryl groups on proteins and can alter protein conformation and activity (Needleman 2004). Pb can form protein complexes in the kidney’s proximal tube, which lead to poor blood pressure regulation and cardiovascular problems (Goyer 1989). Pb can also compete with or replace other divalent cations including calcium and iron. In the central nervous system, rodent model research indicates lead exposure reduces synaptic plasticity and hippocampal long-term potentiation (Toscano and Guilarte 2005). It also results in a cascade of molecular changes including reduced cAMP and protein kinase A activity, altered MAPK signaling, and disruption of CREB phosphorylation (Toscano and Guilarte 2005). Pb and Hcy both independently exert toxicity in part by binding to free sulfhydryl groups and damaging proteins.

Exposure to Pb is linked to numerous diseases and negative health outcomes throughout the lifespan. In late life, lead exposure, even at low levels experienced in the community, is related to CVD, as measured through hypertension (Cheng et al. 2001), heart rate variability (Park et al. 2006), and clinical CVD outcomes (Navas-Acien et al. 2007). Lead exposure is also associated with poor cognition and cognitive decline in aging populations (Shih et al. 2007; Weuve et al. 2009).

Lead Exposure and Homocysteine

Previous studies have demonstrated a relationship between Pb and Hcy at single time points. In the Baltimore Memory Study, blood Pb and plasma Hcy were significantly correlated with each other (Pearson’s unadjusted r=0.27) (Schafer et al. 2005). Similarly, in Pakistan, a cross-sectional survey of persons aged 18-60 years, an increase of 1 μg/dL log blood Pb was associated with an increase of 0.09 μmol/L log Hcy after multivariable adjustment (Yakub and Iqbal

43 2010). In a third cross-sectional study, this time in Vietnam and Singapore, an increase of 1μg/dl log blood lead was associated with an increase of 0.04μmol/l log Hcy among occupationally exposed 449 workers (mean age=39; mean blood Pb= 22.7 μg/dl) (Chia et al. 2007).

One reason that lead exposure and elevated Hcy are associated with many of the same negative health outcomes is that lead exposure elevates Hcy. For example, in the 1999-2002 National Health and Nutrition Examination Survey (NHANES), Hcy was associated with peripheral arterial disease (PAD) (OR of PAD in the highest quintile of Hcy relative to the lowest was 1.92, p-trend=0.004), however the association disappeared following adjustment for blood lead and calcium levels as well as kidney function (OR=0.89, p-trend=0.87) (Guallar et al. 2006). This research suggests the Hcy may be a marker of exposure to other toxicants, rather than the toxic agent itself.

Pb and Hcy levels may be mechanistically linked. Several proteins in the homocysteine processing one-carbon metabolism cycle (FIGURE 2.1) contain sulfhydryl groups that may be potential reaction sites for lead (Schafer et al. 2005). Cystathionine β-synthase (CBS) catalyzes the breakdown of Hcy into cysteine. CBS has two sulfhydryl groups that may be subject to reaction with lead, potentially interfering with the enzyme’s ability to transsulfurate Hcy into cysteine (Schafer et al. 2005). In addition, Hcy contains a sulfhydrl group and lead may directly inhibit its metabolism. There is also evidence for the protective effect of methionine on lead toxicity (Chen et al. 2011). It is hypothesized that Pb could directly influence the levels of Hcy in the body.

The association reported between lead and Hcy is plausible and supported by compelling data, but it requires further epidemiologic investigation in an additional population. Thus far, only one study has tested the Pb-Hcy relationship in a community-exposed population of older adults. There are currently no studies evaluating the relevant window of Pb exposure’s effect on Hcy. Studies of cumulative Pb exposure and Hcy are needed to fill this gap. In

44 addition, there are no data indicating whether the association of Pb exposure with Hcy can be mitigated (or worsened) with intake of B vitamins.

Objectives

The goal of this study was to evaluate the relationship of lead exposure to plasma Hcy in a population of community-exposed older men. We tested the hypothesis that recent exposure to Pb (meausured by blood Pb) is associated with concurrent Hcy levels. Next, we evaluated whether people who have changes in their blood Pb over time also have corresponding changes in their Hcy levels. We also tested whether those with a diet low in methyl donors (folate, vitamin B6, and vitamin B12) are more susceptible to the effects of Pb exposure on Hcy. Finally we evaluated the relationship of cumulative Pb dose (tibia and patella bone Pb) and Hcy in comparison to acute/recent Pb exposure.

METHODS

Study Population

In 1963, 2,280 men in the greater Boston area between the ages of 21 and 80, and representing a range of educational and occupational backgrounds, were enrolled in the Veteran’s Affairs Normative Aging Study (NAS) (Bell et al. 1972). All participants were free of disease at the onset of the study and participated in health assessments every three to five years that expanded in scope over time. Blood lead measurements began in 1977 and bone lead measurements began in 1991. Homocysteine was first measured in 1993. At each study visit, age, smoking status, medication use, physical activity, and dietary intake were assessed.

In the study of Hcy, up to six repeated measures of Hcy from 1,080 men for a total of 2,941 observations were available. Bone lead data was missing for 301 individuals and there were a total of 779 subjects across 2,252 observations that had bone lead data and Hcy measures. Approximately three years after the bone lead assessment, the first Hcy was measured. Concurrent blood lead data

45 was missing from 23 participants and there were 1,056 individuals with 2,301 observations of Hcy and blood Pb. In an additional analysis of the changes in blood Pb and Hcy, a subset of the main dataset was used. At least two repeated measures of blood lead and concurrent Hcy were available from 747 men for a total of 1,830 repeated observations. The majority of blood lead measurements (>99%) were made within 30 days of Hcy measurements. The Human Subjects Institutional Review Boards at the Harvard School of Public Health, the Department of Veterans Affairs Outpatient Clinic in Boston, the Brigham and Women’s Hospital, and the University of Michigan Medical School approved this study.

Plasma Homocysteine Measures

Fasting blood plasma was collected during each clinic visit and frozen at negative 80ºC. Samples were analyzed at the Jean Mayer US Department of Agriculture Human Nutrition Research Center on Aging. Total Hcy was measured via fluorescence detection with high-performance liquid chromatography (HPLC) (Araki and Sako 1987). Hcy detection methods have been previously described in detail (Tucker et al. 2005). Plasma folate levels below 3 ng/mL were considered low (LSRO 1984). Plasma B6 levels at 30nmol/L were considered adequate (US RDA, NIH Office of Dietary Supplements). Plasma B12 below 250 pg/mL was considered low (US RDA, NIH Office of Dietary Supplements). Participants with low nutrient status were included in the analysis.

Lead Exposure Measures

Lead levels can be measured in various tissue types across the body to reflect different exposure times. Blood lead levels indicate recent exposure (with a half-life of approximately 30 days) (Hu et al. 1998). Blood lead levels included in this analysis were measured up to six times between 1993 and 2011 via graphite furnace atomic absorption with Zeeman background correction.

46 Bone lead was measured in two bone types (tibia and patella) through Cd109 K-shell x-ray fluorescent (KXRF) spectroscopy using methods previously described (Hu et al. 1996). The half-life for lead in cortical tibia bone is estimated at 48.6 years in the NAS (Wilker et al. 2011). The trabecular patella bone has a shorter half-life for lead (between 10-15 years) (Hu et al. 1998). The primary statistical analyses were performed using tibia Pb, while patella Pb was used as a secondary sensitivity analysis.

Dietary Measures

Annual average diet was assessed with the semi-quantitative Willett Food Frequency Questionnaire using methods described previously (van de Rest et al. 2009). Before each study visit, participants were mailed the questionnaire. Their responses were checked for completeness at the study visit. The questionnaire assesses frequency of consumption of 126 items on a scale ranging from never to ≥2 times per day. Three one-carbon metabolism dietary factors were measured without supplements and were used in the current study (folate, vitamin B6, and vitamin B12). Nutrient quantification based on FFQs administered after August, 1997 were adjusted to consider folate fortification in the US. Total calorie intake-adjusted nutrient residuals (Willett et al. 1997) (Willett et al. 1997) were calculated for each individual and included in the models.

Statistical Analysis

All analyses were performed in R Statistical Software (version R 2.15.0). Univariate descriptive statistics were calculated for each variable. We also estimated bivariate associations between Hcy, and the three lead exposure measures and additional covariates were also calculated.

We assessed the associations between blood Pb and concurrent Hcy using multivariable-adjusted linear models. Because the distribution of Hcy levels is skewed, we natural log-transformed the variable Hcy. Thus, the exponentiated parameter estimates from these models are directly interpreted as

47 the percentage difference in Hcy per unit increment in predictor. We adjusted these analyses for several sets of covariates. All analyses were adjusted for the following “core covariates”: age, education, smoking status, alcohol consumption, and body mass index (BMI) (Schafer et al. 2005). The following plasma levels of one-carbon metabolism related compounds were considered the “plasma sensitivity covariates”: plasma PLP (a measure of vitamin B6), vitamin B12, and folate. Finally, dietary measures were categorized as the “dietary sensitivity covariates”: intakes of total energy and calorie adjusted residuals for vitamin B6, vitamin B12, and folate. We tested for non-linear trends using generalized additive models with penalized splines in the mgcv R package.

Using the repeated measures of Hcy and linear mixed effects models with random intercepts, we compared rates of change in Hcy (as percentage change from baseline Hcy) by level of lead biomarker. We modeled interaction terms between time and baseline age to capture the trajectories of Hcy over the follow- up period. Several error covariance matrices were considered, but they did not improve model fit, so no within-subject correlation was used. All models were adjusted for Core Covariates, Plasma Sensitivity Covariates, and Dietary Sensitivity Covariates, similar to the concurrent exposure models. In addition, all models included a cross-product term between time and baseline age. The inclusion of other time-covariate cross-products did not change the results.

We used “change-change” models to assess whether change in Pb level over time was associated with concomitant changes in Hcy. In these models, we restricted the analyses to individuals with at least two measures of Pb and Hcy (747 individuals, 1,830 observations). We regressed change in blood Pb on change in Hcy, adjusting for the core covariates.

To assess the potential modification of the association between Pb and Hcy by key nutrients, we fit models stratified by median plasma or dietary levels of folate and vitamins B6 and B12. We used methods by Payton et al. (Payton et al. 2003) to test for differences in the effect of lead on Hcy between the high and low nutrient strata. This permits other covariate-Hcy associations to vary across

48 the strata, associations that may influence the stratum-specific Pb association. In addition, however, we fit models, incorporating all observations that included a cross-product term between each Pb marker measure and nutrient measure.

Additional analyses were performed to determine whether the effects of cumulative lead dose were mediated through current blood lead levels. Our secondary aim was to determine whether lead-associated changes in Hcy were associated with recent exposure measured by blood lead or cumulative dose measured by tibia lead. We used mixed effects core covariate models containing tibia Pb, both tibia Pb and blood Pb, and blood Pb alone as a basic mediation analysis.

RESULTS

The univariate descriptive statistics for subjects with blood Pb levels are presented in Table 1. These values are similar to those for the population with tibia bone lead measures. The geometric mean Hcy level in this population was 10.2 nmol/ml (GSD= 1.3) (TABLE 2.1). Mean (SD) blood, tibia bone, and patella bone lead levels were 4.4(2.5) µg/dl, 21(13) µg/g, and 29.9(19.5) µg/g, respectively. Similar to the general US population at this time, blood lead levels decreased across study visits. In general, this population was elderly (age 70.9 (7.3)), overweight (BMI 28(4)), educated (64.2% with some post-high school education), moderate-heavy drinkers (20.6% with more than 2 drinks per day), and former smokers (66.2%).

At visit one, log(Hcy) was positively correlated with blood Pb, tibia bone Pb, and patella bone Pb (respectively Pearson’s r=0.28 p=2x10-16; r=0.15 p=2x10-5; r=0.16, p=3x10-5) (Supplemental Figure 2.1). The bivariate associations between Hcy and other covariates before and after age-adjustment at study visit 1 are listed in Table 2.2. Hcy was positively associated with increasing linear trend in categories of blood Pb, tibia bone Pb, patella Pb, age, alcohol consumption, and smoking status. Hcy was inversely associated with plasma PLP (B6), B12, folate and dietary B6 levels. Similar associations were

49 observed at the other study visits (data not shown). A notable exception is that smoking was only associated with Hcy at study visit one. Similar results are also observed in the population with tibia bone Pb levels (n=774) (data not shown). The bivariate associations between covariates and blood Pb as well as tibia bone Pb are also presented in Table 2.2. Blood Pb was positively associated with age and alcohol intake. Blood Pb was negatively associated with increased education, plasma B6, plasma B12, plasma folate, dietary B6, and dietary B12. Tibia bone Pb was positively associated with smoking status (current smokers with higher exposure) and tibia bone Pb was negatively associated with increased education, plasma B6, plasma folate, and dietary B6.

After adjustment for age, education, alcohol, smoking, and BMI (Core Covariates) an IQR increment in blood Pb (3 ug/dl) at visit 1 was associated with an 8.0 percent increase in Hcy (95% CI: 6.0-10.0). By comparison, a 5-year increment in age was associated with a 3.2 percent increase in Hcy (95% CI=2.0- 4.4) (FIGURE 2.2A). The percent difference in Hcy at visit 1 between never smokers and current smokers was 14.0 (95% CI: 5.6-23.1). The difference in Hcy between never smokers and former smokers was not statistically significant. Next we added to the model plasma factors that influence Hcy metabolism. After further adjustment for the plasma sensitivity covariates (plasma PLP, B12 and folate), an IQR increment in blood Pb was associated with 6.4% increase in Hcy (95% CI: 4.5-8.3), while after further adjustment for the dietary sensitivity covariates (core covariates plus total energy consumption, dietary B6, B12, and folate), an IQR increment in blood Pb was associated with 5.6 in Hcy (95% CI: 1.4-9.9). These results were consistent across the 4 study visits (FIGURE 2.2A).

Results from the mixed effects models of the repeated blood Pb and Hcy measures were consistent with those from the models based on single visits (TABLE 2.3). Plasma B6, plasma B12, plasma folate, and dietary B6 were significant negative predictors of log(Hcy). In the core model, a 3 ug/dl (IQR) increase in blood Pb is associated with 6.0% increase in Hcy (95% CI: 4.6-7.5).

50 The adverse association between blood Pb and Hcy was stronger among men whose dietary intake of vitamin plasma B6, B12, or folate fell below the median B6: 2.24 mg/day, B12: 6.185 mcg/day, folate: 343 mcg/day) (FIGURE 2.4). Results were similar albeit weaker for the plasma vitamin measures. None of the blood lead estimates for men with the higher vitamin levels was significantly different from the corresponding estimates for men with the lower vitamin levels. Data was stratified by median plasma level (B6: 64.1 nmol/L, B12: 447.5 pg/mL, folate: 11.4 ng/mL) or median dietary levels (B6: 2.24 mg, B12: 6.185 mcg, folate: 343 mcg). However, when we modeled the interaction between blood Pb and the nutrients using cross-product terms, the interactions corresponding to dietary B6, dietary folate, and plasma folate were statistically significant (p-values for interaction: 9.6x10-5, 0.0016, and 0.024 respectively), and the interaction corresponding to dietary B12 was borderline significant (p- value for interaction: 0.051).

Among the 747 men with two or more measures of blood Pb and Hcy, blood Pb levels generally declined over time (mean) but this decline was variable (SD or range). Changes in Hcy generally “tracked” changes in blood Pb, with a 1-μg/dL drop in blood Pb corresponding to a 5% drop in Hcy, but this association was not statistically significant (Supplemental Tables 2.1 and 2.2).

Higher cumulative exposure to Pb, as measured by tibia bone lead concentration, was also significantly associated with increased plasma Hcy. Using all four study visits, tibia bone Pb was a significant predictor of log(Hcy) (p- value =0.0012) (TABLE 2.4). Similarly, at visit 1 alone, tibia Pb predicted log(Hcy) (p-value=0.016). Next we looked at each visit individually using linear regression with three sets of covariates. Similar to the blood lead analysis, after adjusting for the Core Covariates (age, education, alcohol, smoking, and BMI) an IQR change in tibia bone Pb (14 ug/g) at visit 1 was associated with a 3.6 percent increase in Hcy (95% CI: 1.4-6.0) (FIGURE 2.2B). After further adjustment for plasma PLP, B12 and folate, an IQR change in tibia Pb was associated with 2.4 percent increase in Hcy (95% CI: 0.2-4.6). Finally in the

51 dietary sensitivity model, an IQR change in tibia Pb was associated with 2.5% increase in Hcy (95% CI: -0.3-5.4). Similar results were observed across the four study visits.

While cumulative exposure to lead was associated with consistently higher plasma Hcy over the course of follow-up, it was not associated with the degree to which plasma Hcy changed over time (TABLE 2.5). On average, Hcy levels increased by about 1.2% per year of follow-up. An IQR increment in tibia Pb (14 µg/g) corresponded to a negligible 0.02% slower percentage change in Hcy over time (95% CI: XX).

There were no significant differences in the concurrent effect of bone Pb on Hcy between persons with low and high plasma or dietary B vitamins or folate. In a subset of individuals (n=221) with at least three repeated visits with dietary information (n=691 observations), we tested for a long term dietary influence on Hcy. We averaged dietary intake of folate and vitamins B6 and B12 over three- four visits and used mixed effects models to test for dietary associations with Hcy. In this smaller set of individuals, tibia Pb was no longer a significant predictor of Hcy and we did not observe a long term dietary trend (Supplemental Table 2.3).

To determine whether the association between cumulative exposure to Pb and Hcy was mediated by current exposure to Pb, we further adjusted the analyses of tibia Pb for blood Pb as a predictor (TABLE 2.6). Adding blood Pb to the model removed the association between tibia Pb and Hcy. We see similar results when we restrict the dataset to just visit one and compare the linear multivariate models containing (1) core covariates and tibia Pb, (2) core covariates, tibia and blood Pb, and (3) core covariates and blood Pb (Supplemental Table 2.4). Tibia Pb is an significant predictor of Hcy at visit one, but it drops out of the model with the addition of blood Pb. This suggests that the association between tibia Pb and Hcy may be mediated by blood Pb.

DISCUSSION

52 This cohort study examined the association between Pb exposure and circulating Hcy levels in a population of elderly community-dwelling men in the greater Boston area. Increased current blood Pb exposure was associated with greater Hcy levels. These effects were stronger in individuals with low dietary vitamins B6, B12, and folate as well as plasma folate. Greater cumulative exposure to Pb, measured in the tibia and patella bones, was associated with higher levels of Hcy. Further analyses revealed that these results may be mediated through current blood Pb and that Hcy responds to acute Pb dose.

The present finding that Pb exposure is associated with Hcy levels is consistent with prior epidemiological studies. The Baltimore Memory Study examined the cross-sectional association between Pb and Hcy in a population of 1,037 adults (mean age= 59) with similar blood Pb and Hcy to our study. The analysis observed a 1.0 ug/dL increase in blood lead was associated with a 0.43 μmol/L increase in Hcy in males, but no association was observed with tibia Pb (Schafer et al. 2005). Among 872 adults in Pakistan age 18-60, with mean blood Pb 11.65± 5.5 μg/dl, an increase of 1 μg/dl log blood Pb was cross-sectionally associated with an increase of 0.09 μmol/l log Hcy (Yakub and Iqbal 2010). In a group of 276 workers occupationally exposed to Pb in Vietnam, blood lead and Hcy were associated (Pearson’s correlation=0.255, p<0.01) (Chia et al. 2007). Our research has confirmed the association between lead exposure and Hcy demonstrated in the above studies. However, these studies did not examine longitudinal change in Hcy and they did not assess how the association might be influenced by intake of folate and B-vitamins.

The results of this study are plausible given that Pb and Hcy share toxic mechanisms. Both form stable disulfide bonds with protein cysteine residues, potentially altering protein function. In particular, albumin (the dominant protein in blood) Cys34 has a low pKa that readily reacts with metals and Hcy (Carter and Ho 1994). In addition, both are associated with inflammation (Hcy through NF-kB activation of IL-8 and MCP-1 (Perla-Kajan et al. 2007); and Pb through cytokine production (Heo et al. 1996)). Given the independent cell culture and

53 animal model evidence on each toxicant (Pb and Hcy), our epidemiological observations on the associations between Pb and Hcy, with interactions with B- vitamins, are plausible. Studies are needed to examine the health effects of combined exposure to Pb and these vitamins, and the degree to which they are mediated by Hcy.

The current analyses are influenced by a few limitations of the data. First, there may be error in the measurement of Hcy. The time of day of blood draw for Hcy measures was not standardized. In healthy individuals over a 24-hour period, Hcy levels showed a daily rhythm characterized by an evening peak and nighttime low (Bonsch et al. 2007). Whole blood genomic DNA methylation varies throughout the day and is inversely correlated with Hcy levels (Bonsch et al. 2007). Unstandardized Hcy collection times in our study would have biased our results towards the null hypothesis of no association between Hcy and Pb. In addition, plasma measures of Hcy assess the pool of Hcy released after reduction of all disulfide bonds in the sample. Total Hcy does not include homocysteine thiolacetone (a product of misincorporation of Hcy into proteins and subsequent error-editing) or Hcy bound to protein by an amide bond (Perla- Kajan et al. 2007). These Hcy groups are potentially toxic and may not be correlated with plasma concentrations of Hcy (Perla-Kajan et al. 2007).

Another potential weakness of this study is the limited power to detect interactions. This problem is exacerbated in our study due to missing nutrient data and smaller sample sizes in dietary models. The change-change analyses also have limited power as participants were lost to follow-up over time.

Epidemiologic studies that identify mediators linking exposure and outcome strengthen the model’s biological plausibility and potential causal relationship (Hafeman 2011). In this study, we measured blood Pb as a mediator of tibia Pb. Our results are based on the assumption that there are no unmeasured confounders of the causal effect of the mediator (blood Pb) on the outcome (Hcy) (Cole and Hernan 2002). With these caveats in mind, our analysis incorporating blood Pb and tibia Pb suggest that the influence of Pb exposure on

54 Hcy is fairly immediate and thus that recent exposure to Pb (measured in blood) may be driving the association between Pb and Hcy. Despite cumulative exposure, interventions that lower acute blood Pb levels may be an effective strategy to lower Hcy.

The current study is strengthened by the use of repeated measures of Hcy and blood Pb as well as repeated measures in Hcy with baseline bone Pb. This allows us to look at longitudinal changes in Hcy with Pb exposure, not simply cross-sectional associations. This study is the first to examine the Pb-Hcy relationship while examining plausible dietary interactions, namely folate and vitamins B6 and B12.

Since Pb exposure is related to elevated Hcy levels, toxicology studies are needed to determine the mechanism and potential reversibility of Pb and disturbed Hcy metabolism. Based on the current study, dietary intervention with folate and vitamins B6 and B12 may be a potential option to remediate elevated Hcy high Pb exposed individuals.

In the 1999-2002 US National Health and Nutritional Examination Study (NHANES), Hcy was cross-sectionally associated with peripheral arterial disease (PAD), but adjustment with blood Pb, Cd, smoking, and glomerular filtration rate removed the Hcy association (Guallar et al. 2006). This suggests that Pb and Hcy levels are related and are associated with cardiovascular effects. Future research may test this association and the association with cognitive decline in cohort studies.

In conclusion, we report a significant association between blood, patella, and tibia Pb levels with higher levels of plasma Hcy in a group of older men. The association corresponding to blood Pb was strongest, suggesting that circulating lead may influence circulating Hcy through its metabolism, even at very low levels of exposure. The effects of chronic lead exposure are also supported by these results.

55 Diet may modify the association between blood Pb and Hcy. The adverse effect of blood lead on Hcy may be worse in the presence of low folate, vitamin B6, and vitamin B12 intake (equivalently: the adverse effect of low folate intake on Hcy may be worse in the presence of lead, even for very modest levels of lead exposure). Altered Hcy levels, such as those observed in the range here, may increase risk for cardiovascular and neurodegenerative disease and measures should be taken to reduce blood lead levels and improve dietary methyl donor diet status.

TABLES Table 2.1. Univariate Statistics: Characteristics of individuals with complete Hcy and blood Pb (2234 obs, 1048 individuals). Mean (S.D.), except where noted.

All Visits Visit 1 Visit 2 Visit 3 Visit 4 1056 individuals N 1056 747 400 98 (2301 observations) N Missing N N N N Parameter Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) (%) Miss(%) Miss(%) Miss(%) Miss(%) Homocysteine (μmol/l)* 10.2(1.3) 0(0) 10.1(1.3) 0(0) 10.2(1.3) 0(0) 10.5(1.3) 0(0) 10.8(1.3) 0(0) Blood Pb (μg/dl) 4.4(2.5) 0(0) 4.9(2.7) 0(0) 4.2(2.3) 0(0) 3.6(2.2) 0(0) 3.4(2.2) 0(0) Age (years) 70.9(7.3) 0(0) 69(7.4) 0(0) 71.3(6.9) 0(0) 74(6.5) 0(0) 76.4(5.9) 0(0) Education < HS [n(%)] 157(6.8) 80(7.6) 48(6.4) 24(6.0) 5(5.1) HS [n(%)] 668(29) 303(28.7) 219(29.3) 118(29.5) 28(28.6) 0(0) 0(0) 0(0) 0(0) 0(0) Some college [n(%)] 630(27.4) 288(27.3) 204(27.3) 107(26.8) 31(31.6) College [n(%)] 449(19.5) 202(19.1) 151(20.2) 79(19.8) 17(17.3)

57 > College [n(%)] 397(17.3) 183(17.3) 125(16.9) 72(18.0) 17(17.3) Smoking Status Never [n(%)] 663(28.8) 296(28.0) 211(28.2) 114(28.5) 42(42.9) 0(0) 0(0) 0(0) 0(0) 0(0) Former [n(%)] 1524(66.2) 698(66.1) 502(67.2) 272(68.0) 52(53.1) Current [n(%)] 114(5) 62(5.9) 34(4.6) 14(3.5) 4(4.1) Alcohol Consumption ≤ 2 drinks/day [n(%)] 1827(79.4) 0(0) 834(79.0) 0(0) 594(49.5) 0(0) 320(80.0) 0(0) 79(80.6) 0(0) > 2 drinks/day [n(%)] 474(20.6) 222(21.0) 153(20.5) 80(20.0) 19(19.4) BMI 28(4) 0(0) 28(3.9) 0(0) 28.2(4) 0(0) 27.9(4.2) 0(0) 27.5(3.7) 0(0) Tibia Pb (μg/g) 21.4(13.5) 285(27) Patella Pb (μg/g) 30.6(20.1) 289(27.4) Plasma B6 (nmol/l) 93.8(92.5) 27(1.2) 87.5(87.2) 12(1.1) 97(93.7) 9(1.2) 99.2(93.5) 5(1.2) 114.9(125.5) 1(1) Plasma B12 (pg/ml) 490(235.8) 26(1.1) 466.4(222) 21(2) 493.4(244.1) 2(0.3) 519.9(230.5) 3(0.8) 593.4(289.4) 0(0) Plasma Folate (ng/ml) 14.2(11.7) 35(1.5) 11(7) 30(2.8) 14.8(12.9) 2(0.3) 19.4(15.2) 3(0.8) 22.2(14.7) 0(0) Total Energy Consumption 1978.3(643. 1967.4(618. 1985.3(629. 2012.7(730. 1898.9(626. 150(6.5) 70(6.6) 43(5.8) 28(7) 9(9.2) (kcal) 1) 4) 6) 8) 3) Dietary B6 (w/o supp) (mg) 2.4(1) 104(4.5) 2.3(0.9) 49(4.6) 2.4(0.9) 25(3.3) 2.4(1.1) 22(5.5) 2.4(1.1) 8(8.2) Dietary B12 (w/o supp) (mcg) 7.5(4.8) 104(4.5) 7.9(5.5) 49(4.6) 6.9(3.8) 25(3.3) 7.4(4.4) 22(5.5) 8.1(5) 8(8.2) Dietary Folate (w/o supp) 379.1(179) 1034(44.9) 344(150.5) 794(75.2) 356.5(164) 210(28.1) 418.5(195) 22(5.5) 451.2(218.4) 8(8.2) (mcg) * Geometric Mean and Geometric Standard Deviation are reported.

Table 2.2. Bivariate statistics based on visit 1. Characteristics of individuals with complete Hcy and blood Pb (1056 individuals).

Geometric Age-Adjusted Age- Mean Blood Mean Tibia Variable Category N (%) Mean Hcy P-trend Least Squares Adjusted P-trend P-trend Pb (SD) Pb (SD) (GSD) Mean Hcy P-trend ≤ 3 367(34.8) 9.3(1.3) 9.7 2.4(0.8) 16(8.5) Blood Pb (ug/dl) 3 5 342(32.4) 11.1(1.4) 11.6 8(2.4) 27.4(16.4) ≤ 15 267(25.3) 9.5(1.3) 10.0 3.7(2.1) 9.8(4.1) 15 23 256(24.2) 10.7(1.3) 11.1 6.2(3.2) 35.7(13.3) Missing 285(27) 10.2(1.3) 10.6 4.8(2.6) NaN(NA) ≤ 20 255(24.1) 9.7(1.3) 10.1 3.8(2.1) 13.4(7.1) Patella Pb 2034 253(24) 11(1.3) 11.3 6.3(3.2) 32.5(15.6) Missing 290(27.5) 10.1(1.3) 10.6 4.8(2.6) 13.6(12.5)

58 50-65 355(33.6) 9.8(1.4) 10.6 4.6(2.6) 16.5(9.3) Age Tertile 66-71 327(31) 9.9(1.3) 2E-05 10.3 0.8 4.9(2.8) 0.03 21.8(12.3) 0.4 72-97 374(35.4) 10.7(1.3) 10.8 5(2.8) 26(16.3) < HS 80(7.6) 9.8(1.3) 10.1 5.8(3) 30.2(18.6) HS 303(28.7) 10.2(1.4) 10.7 5.2(2.9) 24.7(15.9) Education Some College 288(27.3) 10.1(1.3) 0.9 10.6 0.8 4.8(2.7) 8E-06 20.3(11) 3E-15 College 202(19.1) 9.9(1.3) 10.3 4.5(2.3) 18.9(11.3) > College 183(17.3) 10.2(1.3) 10.7 4.4(2.8) 17(9.4) Never 296(28) 9.8(1.3) 10.2 4.8(3) 20(13.7) Smoking Status Former 698(66.1) 10.1(1.3) 0.005 10.6 0.0002 4.8(2.6) 0.1 22(13.6) 0.002 Current 62(5.9) 11.2(1.3) 12.0 5.8(3.2) 22.1(11.5) No 834(79) 9.9(1.3) 10.3 4.6(2.7) 21.4(13.9) Two Drinks/Day 0.00007 0.000006 2E-07 0.5 Yes 222(21) 10.8(1.4) 11.5 5.7(2.9) 21.2(12.1) < 25 216(20.5) 10.1(1.3) 10.4 4.9(2.8) 21.4(11.5) BMI 25≤x<30 574(54.4) 10.1(1.3) 0.9 10.6 0.5 4.8(2.7) 0.5 21.8(14.3) 0.3 ≥30 266(25.2) 10.1(1.3) 10.6 4.8(2.8) 20.4(13.3) <30 108(10.2) 11.3(1.4) 12.0 5.7(3.4) 27.2(14.6) Plasma B6 ≥30 936(88.6) 10(1.3) 1E-05 10.4 7E-06 4.7(2.6) 5E-04 20.6(13.2) 2E-05 (nmol/L) Missing 12(1.1) 9.5(1.2) 9.7 5.3(2.8) 30.7(12.4) <250 82(7.8) 12.3(1.4) 13.0 5.8(3.2) 22.6(12.6) Plasma B12 ≥ 250 953(90.2) 9.9(1.3) 1E-10 10.3 2E-11 4.8(2.7) 6E-04 21.2(13.5) 2E-01 (pg/mL) Missing 21(2) 10.5(1.4) 11.2 4.9(2.4) 28.2(24.1)

< 3 24(2.3) 13.7(1.5) 15.0 7.2(4.2) 29.1(15.2) Plasma Folate ≥ 3 1002(94.9) 10(1.3) 9E-08 10.4 1E-08 4.8(2.7) 3E-05 21.1(13.2) 4E-03 (ng/mL) Missing 30(2.8) 10.3(1.4) 10.8 4.7(2.2) 28.6(21.5) <2300 739(70) 10.1(1.3) 10.5 4.9(2.8) 21.6(13.7) Total Energy ≥2300 247(23.4) 10.2(1.3) 0.8 10.7 0.7 4.7(2.8) 0.5 20.8(13.6) 0.6 Consumption Missing 70(6.6) 10.1(1.3) 10.4 4.8(2.4) 21.4(10.5) <1.7 774(73.3) 10(1.3) 10.4 4.7(2.7) 21.1(13.2) Dietary B6 (mg) ≥1.7 233(22.1) 10.6(1.3) 0.006 11.2 0.001 5.3(2.8) 0.007 22.7(15) 0.05 Missing 49(4.6) 10.1(1.4) 10.4 4.8(2.5) 20.2(10.9) <2.4 960(90.9) 10.1(1.3) 10.5 4.8(2.7) 21.3(13.6) Dietary B12 ≥2.4 47(4.5) 10.6(1.4) 0.3 11.3 0.2 5.6(2.9) 0.04 25(14.7) 0.06 (mcg) Missing 49(4.6) 10.1(1.4) 10.4 4.8(2.5) 20.2(10.9) <400 73(6.9) 9.9(1.3) 10.1 4.1(2.2) 19.8(19.6) Dietary Folate ≥400 189(17.9) 9.9(1.3) 0.9 10.2 0.8 4.6(2.5) 0.1 21.1(14.7) 0.2 (mcg) Missing 794(75.2) 10.2(1.3) 10.7 5(2.8) 21.6(12.5)

Table 2.3. Concurrent Pb exposure is associated with plasma homocysteine: Mixed effects model, random intercept only. Equivalent to cross-sectional model taking into account correlated nature of observations from the same individual. Linear

mixed effects model of log(Hcy)t = b0 + b1[blood Pb]t + covariates. Continuous variables have been centered so the intercept is interpretable.

Core Model* Plasma Model** Diet Model***

2301 obs from 1056 indiv 2240 obs from 1033 1241 obs from 779 indiv R2=0.10 R2=0.16 R2=0.13 Beta(SE) p-value Beta(SE) p-value Beta(SE) p-value

(Intercept) 2.30(0.015) 0 2.30(0.014) 0 2.29(0.017) 0 Blood Pb (per µg/dl) 0.0195(0.0023) 4E-17 0.0159(0.0023) 6E-12 0.0197(0.0035) 2E-08 Age 0.0091(0.00084) 2E-26 0.0105(0.00085) 2E-33 0.0118(0.0012) 3E-22 Education -0.0127(0.013) 0.3 -0.0157(0.013) 0.2 0.00255(0.017) 0.9 (reference>hs)

60 Alcohol 0.0576(0.015) 2E-04 0.0572(0.015) 0.0002 0.0713(0.02) 0.0005 Consumption Former Smoker 0.0176(0.017) 0.3 0.0148(0.016) 0.4 0.0123(0.019) 0.5 Never Smoker 0.0690(0.032) 0.03 0.0646(0.031) 0.04 -0.0077(0.042) 0.9 BMI 0.00191(0.0018) 0.3 0.00129(0.0017) 0.5 0.00326(0.0021) 0.1 Plasma B6 -0.000256(6e-05) 2E-05

Plasma B12 -0.000186(2.4e-05) 5E-15

Plasma Folate -0.00115(0.00043) 0.008

Dietary B6 Residual -0.0310(0.014) 0.03

Dietary B12 Residual -0.00308(0.0021) 0.1

Dietary Folate 3.03e-05(6.4e-05) 0.6 Residual Total Energy Intake -1.64e-05(1.1e-05) 0.1

*Core model adjusts for blood Pb, age, bmi, education, smoking status, and alcohol consumption. **Plasma model adjusts for core model covariates and plasma B6 (PLP), B12, and folate. ***Diet model adjusts for core model covariates and total energy adjusted dietary FFQ vitamin B6, vitamin B12, and folate.

Table 2.4. Cumulative exposure is associated with plasma homocysteine: Mixed effects model, random intercept only. Equivalent to cross-sectional model taking into account correlated nature of observations from the same individual.

Core Model

Visits 1-4 Visit 1

2158 observations from 777 individuals 777 obs from 777 individuals R2=0.6 R2=0.061 β(SE) p-value β(SE) p-value

(Intercept) 2.32(0.016) 0 2.29(0.02) 0 Tibia Pb 0.00215(0.00066) 0.001 0.00256(0.00081) 0.002 Age 0.00921(0.00094) 4E-22 0.00526(0.0015) 0.0007 High School -0.0148(0.015) 0.3 -0.036(0.022) 0.09 Two Drinks/Day 0.0675(0.017) 8 E-05 0.0817(0.025) 0.001

Former Smoker 0.00934(0.019) 0.6 0.0291(0.022) 0.2 Current Smoker 0.0422(0.035) 0.2 0.143(0.047) 0.002 BMI 0.00113(0.002) 0.6 0.00278(0.0026) 0.3

Table 2.5. Longitudinal mixed effects models: Log(hcy) is outcome and tibia pb is main predictor. Continuous covariates have been mean adjusted so that the intercept can be interpreted as the log(hcy) at the mean of those covariates and when dummy variables =0. Four visits of tibia Pb and Hcy have been used. Random intercept and slope.

Core Model* Plasma Sensitivity Model** Diet Sensitivity Model*** n= 2158, R2=0.0811 n= 2106, R2=0.114 n= 1328, R2=0.098 β (SE) p-value β (SE) p-value β (SE) p-value Intercept 2.28(0.017) 0 2.25(0.019) 0 2.24(0.022) 0 Tibia Pb 0.00255(0.00077) 0.0009 0.00254(0.00076) 0.0009 0.00214(0.001) 0.04 Time Since Baseline 0.0123(0.0014) 2E-18 0.0149(0.0015) 1E-23 0.0178(0.0022) 6E-15 Baseline Age 0.0049(0.0015) 0.0009 0.005(0.0014) 0.0006 0.00665(0.0021) 0.001 Education -0.0186(0.014) 0.2 -0.0211(0.014) 0.1 -0.00374(0.018) 0.8 Alcohol Consumption 0.0701(0.017) 4E-05 0.0684(0.017) 6E-05 0.0748(0.022) 0.0008 Former Smoker 0.00509(0.019) 0.8 0.00736(0.018) 0.7 0.0034(0.021) 0.9 62 Current Smoker 0.0301(0.036) 0.4 0.031(0.035) 0.4 -0.00147(0.045) 0.9 BMI 0.00131(0.002) 0.5 0.000636(0.002) 0.8 0.00288(0.0024) 0.2 Plasma B6 -0.000165(5.8e-05) 0.004 Plasma B12 -6.4e-05(1.8e-05) 0.0005 Plasma Folate -0.00159(4e-04) 7E-05 Dietary B6 -0.0284(0.014) 0.04 Dietary B12 -0.000564(0.0018) 0.8 Dietary Folate 7.84e-05(6.4e-05) 0.2 Total Energy Intake -3.03e-05(1.1e-05) 0.005 Baseline Age * Time 0.000881(0.00022) 8E-05 0.000901(0.00022) 5E-05 0.000675(0.00031) 0.03 Tibia Pb * Time -1.72e-05(0.00011) 0.9 -3.24e-05(0.00011) 0.8 3.77e-05(0.00015) 0.8 *Core model adjusts for tibia Pb, baseline age, bmi, education, smoking status, alcohol consumption, time since baseline, baseline age*time since baseline, and tibia Pb*time since baseline. **Plasma model adjusts for core model covariates and plasma B6 (PLP), B12, and folate. ***Diet model adjusts for core model covariates and total energy adjusted dietary FFQ vitamin B6, vitamin B12, and folate.

Table 2.6. Basic mediation analysis. All visits main effect of tibia Pb with and without current blood Pb adjustment. Note: Tibia Pb is no longer significant after adjusting for blood Pb. Research Question: Does cumulative exposure to Pb influence Hcy levels, independent of current exposure to Pb?

Tibia Only Blood and Tibia Blood Only

n=1766 obs, n=771 indiv n=1766 obs, n=771 indiv n=1766 obs, n=771 indiv

R2=0.0711 R2=0.104 R2=0.104

Beta(SE) p-value Beta(SE) p-value Beta(SE) p-value

(Intercept) 2.3(0.017) 0 2.3(0.017) 0 2.3(0.016) 0 Tibia Pb 0.0025(0.00069) 0.0003 0.000832(0.00072) 0.3

Blood Pb 0.0183(0.0028) 4E-11 0.0194(0.0026) 8E-14

Age 0.00715(0.001) 2E-12 0.00886(0.001) 2E-17 0.00922(0.00098) 2E-20

63 Education (>hs -0.0148(0.015) 0.3 -0.0171(0.015) 0.3 -0.014(0.015) 0.4

reference) Alcohol Consumption 0.0794(0.018) 1E-05 0.0682(0.018) 0.0001 0.0676(0.018) 0.0001 Former Smoker 0.00525(0.019) 0.8 0.0101(0.019) 0.6 0.0119(0.019) 0.5 Current Smoker 0.0606(0.037) 0.1 0.0505(0.036) 0.2 0.0521(0.036) 0.2 BMI 0.000694(0.0021) 0.7 0.00162(0.002) 0.4 0.00172(0.002) 0.4

FIGURES

FIGURE 2.1. One-Carbon Metabolism Pathway. Homocysteine can be elevated in conditions of low folate, low B6, or low B12. The sulfhydryl groups on several proteins, including Cystathionine β-synthase (CBS), in the one-carbon metabolism pathway are potential sites for lead’s interferences. Abbreviations Used: Adenosine (Ado), S-adenosylmethionine (SAM), S-adenosylhomocysteine (SAH), Glutathione (GSH), Glutamate (Glu), Glycine (Gly), Tetrahydrofolate (THF), Methionine Transferase (MT), Betaine-Homocysteine S-Methyltransferase (BHMT), Dimethylglycine (DMG), Methylenetetrahydrofolate Reductase (MTHFR)

A. B.

a s

e Core Model

r a

c Plasma Sensitivity Model

n r

I Diet Sensitivity Model

c Core Model

)

n g

I Plasma Nutrient Model

) g

l Diet Nutrient Model

(

n e

c Visit 1 Visit 2 Visit 3 Visit 4

r e

P Visit 1 Visit 2 Visit 3 Visit 4

Figure 2.2. The adjusted cross-sectional association between Pb exposure and homocysteine. Y-axis: With an IQR increase in Pb exposure (blood: 3 ug/dl; tibia Pb: 14 ug/g), corresponding percent increase in plasma Hcy. X-axis: Visit number. Results from cross-sectional multivariate linear regression models. Core covariates are age, blood Pb, education, smoking status, alcohol status, and BMI. Plasma sensitivity covariates include the core model plus plasma PLP, B12, and folate. Diet sensitivity model includes the core covariates and the total energy consumed and the dietary adjusted residuals for B6, B12, and folate. (A) Blood Pb. (B) Tibia Pb.

Baseline Tibia Pb Predicts Hcy Over Time

. 2

Highest Tibia Quartile >27 ug/g

Tibia Pb 18-27 ug/g 5

. Tibia Pb 12-18 ug/g

2 Lowest Tibia Pb <12 ug/g

)

(

. 2

0 2 4 6 8 10 12 14

Time elapsed from baseline (years)

Figure 2.3. Longitudinal core model. Tibia Pb. Data has been centered on the mean continuous variables so intercept can be interpretable. Tibia Pb exposure quartiles are defined at baseline:

)

3 (

Plasma Measure Dietary Measure

s 1

a * e

r * * *

g n

a Below Median Level of Nutrient h

C Above Median Level of Nutrient

e P

B6 B12 Folate B6 B12 Folate

Figure 2.4. Adjusted association between Pb and homocysteine stratified by nutrient status. Y-axis: Percent change in Hcy with an IQR increase in Pb biomarker. Analyses stratified by level of B6, B12 or folate measured in either plasma or dietary FFQ. Longitudinal mixed effects regression of Pb exposure (either blood or tibia bone) on log(homocysteine). *High group is significantly different from low group (p for interaction <0.05).

REFERENCES

Appel LJ, Miller ER, 3rd, Jee SH, Stolzenberg-Solomon R, Lin PH, Erlinger T, Nadeau MR, Selhub J. 2000. Effect of dietary patterns on serum homocysteine: results of a randomized, controlled feeding study. Circulation 102(8): 852-857.

Araki A, Sako Y. 1987. Determination of free and total homocysteine in human plasma by high-performance liquid chromatography with fluorescence detection. J Chromatogr 422: 43-52.

Bell BA, Rose C, Damon A. 1972. The Normative Aging Study: an interdisciplinary and longitudinal study of health and aging. Aging Human Develop 3: 4-17.

Bonsch D, Hothorn T, Krieglstein C, Koch M, Nehmer C, Lenz B, Reulbach U, Kornhuber J, Bleich S. 2007. Daily variations of homocysteine concentration may influence methylation of DNA in normal healthy individuals. Chronobiol Int 24(2): 315-326.

Carter DC, Ho JX. 1994. Structure of serum albumin. Adv Protein Chem 45: 153- 203.

Chen T, Li YY, Zhang JL, Xu B, Lin Y, Wang CX, Guan WC, Wang YJ, Xu SQ. 2011. Protective effect of C(60) -methionine derivate on lead-exposed human SH-SY5Y neuroblastoma cells. J Appl Toxicol 31(3): 255-261.

Cheng Y, Schwartz J, Sparrow D, Aro A, Weiss ST, Hu H. 2001. Bone lead and blood lead levels in relation to baseline blood pressure and the prospective development of hypertension: the Normative Aging Study. Am J Epidemiol 153(2): 164-171.

Chia SE, Ali SM, Lee BL, Lim GH, Jin S, Dong NV, Tu NT, Ong CN, Chia KS. 2007. Association of blood lead and homocysteine levels among lead exposed subjects in Vietnam and Singapore. Occup Environ Med 64(10): 688-693.

Clarke R, Birks J, Nexo E, Ueland PM, Schneede J, Scott J, Molloy A, Evans JG. 2007. Low vitamin B-12 status and risk of cognitive decline in older adults. Am J Clin Nutr 86(5): 1384-1391.

Cole SR, Hernan MA. 2002. Fallibility in estimating direct effects. Int J Epidemiol 31(1): 163-165.

Collaboration HS. 2002. Homocysteine and risk of ischemic heart disease and stroke: a meta-analysis. JAMA 288(16): 2015-2022.

Goyer RA. 1989. Mechanisms of lead and cadmium nephrotoxicity. Toxicol Lett 46(1-3): 153-162.

Guallar E, Silbergeld EK, Navas-Acien A, Malhotra S, Astor BC, Sharrett AR, Schwartz BS. 2006. Confounding of the relation between homocysteine and peripheral arterial disease by lead, cadmium, and renal function. Am J Epidemiol 163(8): 700-708.

Haan MN, Miller JW, Aiello AE, Whitmer RA, Jagust WJ, Mungas DM, Allen LH, Green R. 2007. Homocysteine, B vitamins, and the incidence of dementia and cognitive impairment: results from the Sacramento Area Latino Study on Aging. Am J Clin Nutr 85(2): 511-517.

Hafeman DM. 2011. Confounding of indirect effects: a sensitivity analysis exploring the range of bias due to a cause common to both the mediator and the outcome. Am J Epidemiol 174(6): 710-717.

Heo Y, Parsons PJ, Lawrence DA. 1996. Lead differentially modifies cytokine production in vitro and in vivo. Toxicol Appl Pharmacol 138(1): 149-157.

Hu H, Payton M, Korrick S, Aro A, Sparrow D, Weiss ST, Rotnitzky A. 1996. Determinants of bone and blood lead levels among community-exposed middle-aged to elderly men. The normative aging study. Am J Epidemiol 144(8): 749-759.

Hu H, Rabinowitz M, Smith D. 1998. Bone lead as a biological marker in epidemiologic studies of chronic toxicity: conceptual paradigms. Environ Health Perspect 106(1): 1-8.

Jocelyn P. 1972. Biochemistry of the SH group: the occurrence, chemical properties, metabolism and biological function of thiols and disulphides: Academic Press.

Kim JM, Stewart R, Kim SW, Shin IS, Yang SJ, Shin HY, Yoon JS. 2008. Changes in folate, vitamin B12 and homocysteine associated with incident dementia. J Neurol Neurosurg Psychiatry 79(8): 864-868.

Krumdieck CL, Prince CW. 2000. Mechanisms of homocysteine toxicity on connective tissues: implications for the morbidity of aging. J Nutr 130(2S Suppl): 365S-368S.

LSRO LSRO. 1984. Assessment of folate nutritional status of the US population based on data collected in the second National Helath and Nutrition Examination Survey 1976-1980. Federation of American Societies for Experiemental Biology Bethesda, M.D.: 96.

Mei W, Rong Y, Jinming L, Yongjun L, Hui Z. 2010. Effect of homocysteine interventions on the risk of cardiocerebrovascular events: a meta-analysis of randomised controlled trials. Int J Clin Pract 64(2): 208-215.

Navas-Acien A, Guallar E, Silbergeld EK, Rothenberg SJ. 2007. Lead exposure and cardiovascular disease--a systematic review. Environ Health Perspect 115(3): 472-482.

Needleman H. 2004. Lead poisoning. Annu Rev Med 55: 209-222.

Park SK, Schwartz J, Weisskopf M, Sparrow D, Vokonas PS, Wright RO, Coull B, Nie H, Hu H. 2006. Low-level lead exposure, metabolic syndrome, and heart rate variability: the VA Normative Aging Study. Environ Health Perspect 114(11): 1718-1724.

Payton ME, Greenstone MH, Schenker N. 2003. Overlapping confidence intervals or standard error intervals: what do they mean in terms of statistical significance? J Insect Sci 3: 34.

Perla-Kajan J, Twardowski T, Jakubowski H. 2007. Mechanisms of homocysteine toxicity in humans. Amino Acids 32(4): 561-572.

Schafer JH, Glass TA, Bressler J, Todd AC, Schwartz BS. 2005. Blood lead is a predictor of homocysteine levels in a population-based study of older adults. Environ Health Perspect 113(1): 31-35.

Seshadri S, Beiser A, Selhub J, Jacques PF, Rosenberg IH, D'Agostino RB, Wilson PW, Wolf PA. 2002. Plasma homocysteine as a risk factor for dementia and Alzheimer's disease. N Engl J Med 346(7): 476-483.

Shih RA, Hu H, Weisskopf MG, Schwartz BS. 2007. Cumulative lead dose and cognitive function in adults: a review of studies that measured both blood lead and bone lead. Environ Health Perspect 115(3): 483-492.

Toscano CD, Guilarte TR. 2005. Lead neurotoxicity: from exposure to molecular effects. Brain Res Brain Res Rev 49(3): 529-554.

Tucker KL, Qiao N, Scott T, Rosenberg I, Spiro A, 3rd. 2005. High homocysteine and low B vitamins predict cognitive decline in aging men: the Veterans Affairs Normative Aging Study. Am J Clin Nutr 82(3): 627-635. van de Rest O, Spiro A, 3rd, Krall-Kaye E, Geleijnse JM, de Groot LC, Tucker KL. 2009. Intakes of (n-3) fatty acids and fatty fish are not associated with cognitive performance and 6-year cognitive change in men participating in the Veterans Affairs Normative Aging Study. J Nutr 139(12): 2329-2336.

Wald DS, Law M, Morris JK. 2002. Homocysteine and cardiovascular disease: evidence on causality from a meta-analysis. BMJ 325(7374): 1202.

Weuve J, Korrick SA, Weisskopf MG, Ryan LM, Schwartz J, Nie H, Grodstein F, Hu H. 2009. Cumulative exposure to lead in relation to cognitive function in older women. Environ Health Perspect 117(4): 574-580.

Wilker E, Korrick S, Nie LH, Sparrow D, Vokonas P, Coull B, Wright RO, Schwartz J, Hu H. 2011. Longitudinal changes in bone lead levels: the VA Normative Aging Study. J Occup Environ Med 53(8): 850-855.

Willett WC, Howe GR, Kushi LH. 1997. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr 65(4 Suppl): 1220S-1228S; discussion 1229S-1231S.

Yakub M, Iqbal MP. 2010. Association of blood lead (Pb) and plasma homocysteine: a cross sectional survey in Karachi, Pakistan. PLoS One 5(7): e11706.

SUPPLEMENTAL TABLES

Supplemental Table 2.1. Visits 1-2 all samples, linear regression, split by change in blood Pb n=747

Decrease in Blood No Change Blood Pb Increase in Blood Pb All Observations Pb n=171 n=177 n=399 n=747

p- p- p- Beta(SE) Beta(SE) p-value Beta(SE) Beta(SE) value value value (Intercept) -1.29(0.9) 0.15 -0.42(0.68) 0.83 0.449(0.55) 0.41 -0.0266(0.36) 0.94 delta 0.0578(0.15) 0.7 0.0995(0.11) 0.34 0.0401(0.053) 0.45

bage.c 0.172(0.13) 0.18 0.0326(0.084) 0.7 0.0287(0.065) 0.66 0.0598(0.046) 0.2 as.factor(hs)1 0.0937(0.45) 0.83 0.378(0.41) 0.34 -0.515(0.33) 0.12 -0.138 (0.22) 0.54 as.factor(twodrink)1 -0.312(0.58) 0.59 0.762(0.5) 0.13 -0.147(0.38) 0.7 0.072(0.27) 0.79

72 as.factor(smkcat)2 -0.628(0.49) 0.2 -0.294(0.47) 0.53 -0.285(0.36) 0.42 -0.341(0.24) 0.16

as.factor(smkcat)3 -1.18(1.1) 0.27 -1.62(1) 0.11 -1.39(0.83) 0.095 -1.31(0.55) 0.017 bmi.c -0.0405(0.047) 0.39 0.0123(0.053) 0.82 0.0385(0.042) 0.36 0.00542(0.027) 0.84 timebase 0.0543(0.27) 0.047 0.0844(0.15) 0.56 0.135(0.12) 0.26 0.168(0.086) 0.051 bage.c:timebase -0.025(0.037) 0.51 0.00439(0.023) 0.85 -0.0015(0.017) 0.93 -0.00505(0.013) 0.69

Supplemental Table 2.2: All visits, all samples, mixed effects. n=747 with 1245 observations The average change in Hcy over the average time interval (3.9 years).

Delta Change from Baseline Model Adjacent Change Model N=747 indiv, n=1245 obs N=747 indiv, n=1245 obs R2=0.0482 R2=0.022 Beta(SE) p-value Beta(SE) p-value (Intercept) 0.263(0.23) 0.26 (Intercept) 0.337(0.26) 0.2 delta 0.133(0.042) 0.0015 d.adj.bpb 0.0475(0.045) 0.29 bage.c 0.0125(0.023) 0.59 age.c 0.0469(0.013) 0.00043 as.factor(hs)1 0.0286(0.19) 0.88 as.factor(hs)1 -0.0231(0.18) 0.9 as.factor(twodrink)1 -0.163(0.24) 0.49 as.factor(twodrink)1 0.104(0.21) 0.63

73 as.factor(smkcat)2 -0.557(0.23) 0.013 as.factor(smkcat)2 -0.435(0.19) 0.024

as.factor(smkcat)3 -1.53(0.49) 0.0018 as.factor(smkcat)3 -1.35(0.45) 0.003 bmi.c 0.00321(0.024) 0.89 bmi.c 0.0117(0.021) 0.59 timebase.c 0.134(0.026) 1.9e-7 d.adj.time 0.0722(0.054) 0.18 bage.c*timebase.c 0.00909(0.0036) 0.012

Supplemental Table 2.3: Secondary analysis, long-term dietary trend in tibia model

Core Model: Just Individuals with At Lead Dietary Model on Invididuals With At Least 3 3 Waves of Dietary Information Waves of Dietary Data n=691 obs, n=221 indiv n=691 obs, n=221 indiv R2=0.81 R2=0.088 Variable Beta(SE) p-value Beta(SE) p-value (Intercept) 2.31(0.032) 0 2.31(0.032) 0 tib.c 0.00163(0.0012) 0.16 0.00167(0.0012) 0.15 age.c 0.0131(0.0019) 4.4E-12 0.0135(0.0019) 1.6E-12 as.factor(hs)1 -0.0302(0.028) 0.28 -0.0278(0.028) 0.32 as.factor(twodrink)1 0.0364(0.033) 0.27 0.038(0.033) 0.25

74 as.factor(smkcat)2 0.00264(0.035) 0.94 0.00674(0.036) 0.85 as.factor(smkcat)3 0.0946(0.071) 0.18 0.0869(0.071) 0.22 bmi.c 0.00527(0.0036) 0.14 0.0056(0.0037) 0.13 b6res.lt 0.0304(0.045) 0.5

b12res.lt -0.00972(0.0067) 0.15

- folres.lt 0.49 0.000158(0.00023) calor.c -3.2e-05(1.9e-05) 0.092

***note, when we subset only the people with 3 or 4 dietary measures (n=221 individuals across 691 observations), tibia lead is no longer a significant predictor of log(hcy). None of the long term dietary measures are significant either.

Supplemental Table 2.4: Mediation analysis, visit 1.

Tibia Only Blood and Tibia Blood Only

n=771 n=771 n=771

R2=0.066 R2=0.11 R2=0.11

p- p- β(SE) β(SE) β(SE) p-value value value (Intercept) 2.27(0.02) 0 2.27(0.02) 0 2.26(0.019) 0 Tibia Pb 0.00262(0.00081) 0.001 0.000655(0.00085) 0.4

Blood Pb 0.0241(0.0039) 1E-09 0.0252(0.0036) 9E-12

Age 0.00551(0.0016) 0.0005 0.00587(0.0015) 0.0001 0.00625(0.0014) 2E-05 Education (>hs -0.0408(0.022) 0.06 -0.0495(0.021) 0.02 -0.0463(0.021) 0.03 reference) 75 Alcohol

0.0881(0.025) 0.0004 0.0598(0.025) 0.02 0.0588(0.025) 0.02 Consumption Former Smoker 0.0326(0.023) 0.2 0.0403(0.022) 0.07 0.0419(0.022) 0.06 Current Smoker 0.144(0.047) 0.002 0.134(0.046) 0.004 0.136(0.046) 0.003 BMI 0.00282(0.0027) 0.3 0.00267(0.0026) 0.3 0.00276(0.0026) 0.3

SUPPLEMENTAL FIGURES

Visit 1: Complete Tibia Data Visit 1: Complete Blood Pb Data

)

(

0 . 0 20 40 60 80 100 120 1 0 5 10 15 Tibia Bone Pb (ug/g) Blood Pb (ug/dl)

Supplemental Figure 2.1. Bivariate scatterplots for the unadjusted associations between lead measurement and Hcy at visit 1. Red line is the linear regression line of best fit and the blue lines are the 95% confidence intervals. (A) Tibia lead

(n=774). (B) Patella lead (n=774). (C) Blood lead (n=1048). b

P Restricted to Individuals with Both Plasma and Diet Measures

)

(

- a

l Core Model

n Plasma Nutrient Model i

Diet Nutrient Model

e P Visit 1 Visit 2 Visit 3 Visit 4

Supplemental Figure 2.2. Blood Pb restricted plasma and diet. *Restrict analysis to only individuals with both plasma and dietary measures. (Visit 1 n=774, visit 2 n= 382, visit 3 n= 65, visit 4 n=0)

CHAPTER III Research Chapter 2

Genome-wide DNA Methylation Differences Between Late-Onset Alzheimer’s Disease and Cognitively Normal Controls in the Human Frontal Cortex FROM: Bakulski KM, Dolinoy DC, Sartor MA, Paulson HL, Konen JR, Lieberman AP, Albin RL, Hu H, Rozek LS. 2012. Genome-wide DNA methylation differences between late-onset Alzheimer’s disease and cognitively normal controls in the human frontal cortex. Journal of Alzheimer’s Disease. 29: 571-588, reprinted for educational purposes from IOS Press.

ABSTRACT

Evidence supports a role for epigenetic mechanisms in the pathogenesis of late-onset Alzheimer's disease (LOAD), but little has been done on a genome- wide scale to identify potential sites involved in disease. This study investigates human post-mortem frontal cortex genome-wide DNA methylation profiles between 12 LOAD and 12 cognitively normal age- and gender-matched subjects. Quantitative DNA methylation is determined at 27,578 CpG sites spanning 14,475 genes via the Illumina Infinium HumanMethylation27 BeadArray. Data are analyzed using parallel linear models adjusting for age and gender with empirical Bayes standard error methods. Gene-specific technical and functional validation is performed on an additional 13 matched pair samples, encompassing a wider age range. Analysis reveals 948 CpG sites representing 918 unique genes as potentially associated with LOAD disease status pending confirmation in additional study populations. Across these 948 sites the subtle mean methylation difference between cases and controls is 2.9%. The CpG site with a minimum false discovery rate located in the promoter of the gene Transmembrane Protein 59 (TMEM59) is 7.3% hypomethylated in cases.

Methylation at this site is functionally associated with tissue RNA and protein levels of the TMEM59 gene product. The TMEM59 gene identified from our discovery approach was recently implicated in Amyloid Precursor Protein post- translational processing, supporting a role for epigenetic change in LOAD pathology. This study demonstrates widespread, modest discordant DNA methylation in LOAD-diseased tissue independent from DNA methylation changes with age. Identification of epigenetic biomarkers of LOAD risk may allow for the development of novel diagnostic and therapeutic targets.

Keywords: DNA methylation, Late Onset Alzheimer’s disease, epigenetics, prefrontal cortex

INTRODUCTION

Dementia and Alzheimer’s Disease

Worldwide changes in demography are leading to a rapid increase in the numbers of older adults at risk for dementia. Accordingly, the global prevalence of dementia is expected to quadruple from an estimated 35.6 million cases in 2010 to 115.4 million cases in 2050 (International 2010). The global financial burden of dementia in 2010 was $604 billion (US dollars) including direct medical bills, formal social care, and informal care provided by unpaid caregivers (International 2010).

Alzheimer’s disease (AD), a progressive, fatal neurodegenerative disease, is the most prevalent form of dementia. Less than two percent of AD cases represent early-onset AD (EOAD) (Bird 2005) defined by disease onset prior to age 60 and genetic mutations in amyloid-β precursor protein (AβPP), presenilin-1 (PSEN-1), or presenilin-2 (PSEN-2) genes (Bertram 2009; Hardy 1997). Mutations in these genes dysregulate the AβPP pathway and directly lead to amyloid-β (Aβ) plaque accumulation, a major pathological hallmark of AD.

The remaining vast majority of cases are sporadic, termed Late-Onset Alzheimer’s Disease (LOAD) because they manifest symptoms after age 60. Approximately 60% of LOAD cases carry at least one apolipoprotein ε4 allele (APOE-ε4), while the global population prevalence of the allele is only approximately 22% (Ashford 2004; Kim et al. 2009). Pooled data on LOAD from recent, collaborative, large genome-wide association studies (GWAS) reported the population attributable risk for APOE variants was between 19% and 35% (Ertekin-Taner 2010). GWAS also identified additional LOAD risk alleles (CLU, PICALM, BIN1, CR1, ABCA7, MS4A, EPHA1, CD33, CD2AP) that contribute added risk in population subsets (Harold et al. 2009; Hollingworth et al. 2011; Lambert et al. 2009; Naj et al. 2011). These risk factor genotypes are neither necessary nor sufficient for LOAD development. Twin studies revealed incomplete concordance (Gatz et al. 1997; Nee and Lippa 1999) and variable

age on onset (Li et al. 2002) even among monozygotic pairs, highlighting the complex etiology of LOAD. These observations underscore the likely importance of non-genetic factors in LOAD etiology and spurred recent investigations of the epigenetics of AD.

Epigenetics and Alzheimer’s Disease

Epigenetics is the study of heritable changes in gene expression that occur without changes to the underlying DNA sequence. Methylation (Maunakea et al. 2010) and hydroxymethylation (Jin et al. 2011) at the 5’ site on cytosines in cytosine-guanine (CpG) dinucleotides are important epigenetic modifications associated with gene expression in the human brain. Specific marks distinguish brain regions (Hernandez et al. 2011; Ladd-Acosta et al. 2007) and epigenetic differences in human brain tissues have been associated with such neurological diseases as schizophrenia and bipolar disorder (Mill et al. 2008). Epigenetics is also a mechanism by which environmental exposures can translate to human disease (Dolinoy and Jirtle 2008; Suter and Aagaard-Tillery 2009).

In AD cases lacking highly penetrant genetic susceptibility, the etiology of amyloid dysregulation is not well understood. Altered epigenetic regulation of tau and amyloid processing genes has been observed across multiple brain regions and is a potential mechanism for disease (Barrachina and Ferrer 2009; Tohgi et al. 1999a; Tohgi et al. 1999b). Human post-mortem case-control studies observed global hypomethylation in the entorhinal cortex of AD subjects (Mastroeni et al. 2008) and in the temporal neocortex neuronal nuclei of an AD monozygotic twin relative to their cognitively normal twin (Mastroeni et al. 2009). Evidence for epigenetic involvement in AD pathogenesis spans human studies in various tissues, animal models, and cell culture, and was recently reviewed (2010; Mastroeni et al. 2011; Mill 2011).

Significant transcriptome-wide gene expression differences have been observed between brain tissues of LOAD cases and controls (Loring et al. 2001; Miller et al. 2008). However, previous AD research on DNA methylation as a

regulator of gene expression evaluated DNA methylation at 5’ promoter regions of a few candidate genes selected based on a priori hypotheses about AD molecular mechanisms. The current research provides a semi-unbiased, quantitative, genome-wide discovery of locations of DNA epigenetic differences in human frontal cortex brain tissue between LOAD cases and controls, which allows for identification of novel disease-associated genes. The gene identified in this study that best distinguished cases and controls was technically validated using an additional sensitive and quantitative method of DNA detection. This mark was also validated using a second population of samples. The functional significance of this DNA methylation mark was further confirmed by gene expression and protein quantification assays.

MATERIALS AND METHODS

Sample Acquisition

The NIA funded Michigan Alzheimer’s Disease Center (MADC) (P50AG008671; PI: Sid Gilman) maintains a well-clinically characterized cohort of Alzheimer’s disease and cognitively normal control subjects, many of which agreed to participate in autopsy and donated to the MADC Brain Bank. Upon autopsy, each left hemisphere was fixed in 10% neutral formalin for neuropathological diagnosis. The right hemisphere was sectioned coronally, flash frozen, and archived in MADC freezers at -80ºC. Frozen tissue blocks 0.5 cm3 (50-90 mg) in size were dissected at -20ºC from the mid-frontal gyrus of the frontal lobe and provided for this study. MADC frozen tissues were previously used in high quality expression (Hong et al. 2008; Pan et al. 2007) and proteomic studies (Pan et al. 2007).

Twelve age- and gender- matched pairs of LOAD cases (clinical diagnosis and Braak Score ≥ 4) and controls (clinically confirmed and Braak Score ≤ 2) were used for the genome-wide discovery phase of the project and for gene- specific technical validation. An additional thirteen matched pairs were included in the population validation phase, which included gene-specific DNA

methylation, gene expression, and protein quantification studies. The demographic characteristics of all 50 brains included in this study are described in Table 3.1. Post-mortem intervals in hours for AD cases used in the Discovery Set were as follows: 3, 4, 7, 7, 7.75, 8, 8.75, 9, 11, 12, 14, 24. Post-mortem intervals in hours for controls used in the Discovery Set were as follows: 6, 6, 13.5, 14, 17, 18, 18, 18, 19.3, 20.5, 21.25, 24.5. Gray matter for DNA methylation, expression, and protein analysis was excised from the tissue sample and used in this study and vascular lesions were avoided.

DNA Isolation and APOE Genotyping

DNA was extracted from all 25 matched pair samples using the Promega Maxwell Tissue DNA Kit (Madison, WI) according to manufacturer’s instructions. APOE genotyping was assayed using the Applied Biosystems TaqMan method (Foster City, CA) according to manufacturer’s instructions using the ABI 7900 HT machine (Christensen et al. 2008).

Genome-wide DNA Methylation Discovery

DNA was bisulfite-treated using the Zymo EZ DNA Methylation Kit (Orange, CA) with a modified thermal cycling protocol (98ºC for 10 minutes, 64ºC for 17 hours). Genome-wide DNA methylation was assessed with the Infinium HumanMethylation27 BeadArray (Illumina) performed at the University of Michigan DNA Sequencing Core facility in accordance with manufacturer’s instructions and previously published (Bibikova et al. 2009). Six cases and six control samples were randomly applied to each of two 12-sample arrays to avoid biasing case-control differences by batch effect. BeadArrays were imaged using the Illumina BeadArray Reader. Image processing and intensity data extraction are standard components of the BeadScan software that is associated with the BeadArray Reader. The Illumina BeadStudio Software generated percent methylation estimates (beta values) for each probe set based on Cy3 and Cy5 fluorescence intensities. Data was background normalized and exported for further processing.

Statistical Methods for Bead Array

All statistical analysis was performed with the R Statistical Software (version 2.10.1). CpG sites that failed on 10% of samples were not included in subsequent analyses. Linear models adjusting for age and gender were fit across all CpG sites using the limma package (version 3.2.3). As is standard in microarray analyses, empirical Bayesian variance methods were incorporated to site-specific moderated t-tests (Smyth 2004). The linear model used for each individual CpG site was as follows:

% Methylation = β0 + β1(Case Status) +β2(Age) + β3(Sex)

Top hits by case status were identified by p-values < 0.05 and false discovery rates were calculated. Additional analyses were performed on the top hit. We compared the basic linear model above to a model containing PMI using an F-test. We also compared the R2 goodness of fit of two simple linear regression models containing only either PMI or Case Status as predictors.

Samples were hierarchically clustered by the single linkage method across the top 26 hits by case status. Positional gene set enrichment analysis was performed using Gene Set Enrichment Analysis (GSEA) to determine statistical over-representation of disease specific epigenetic marks within chromosomal cytogenic bands containing at least 15 genes (Subramanian et al. 2005). Enrichment in promoter and 3’UTR regulatory motifs of disease associated genes was determined by GSEA (Xie et al. 2005). Biological processes and molecular functions associated with LOAD gene lists were established using Gene Ontology (Ashburner et al. 2000).

Gene-Specific DNA Methylation Validation

Site-specific methylation technical (of the original 12 Discovery Set matched pairs) and population (of an additional 13 matched pairs with an expanded age range) validation of the top CpG hit was determined by bisulfite- pyrosequencing on the Qiagen Pyromark MD instrument (Valencia, CA). Using

Pyromark Assay Design Software, a custom pyrosequencing assay was designed to include the two CpG sites present on the Illumina array (Target IDs cg01182697 and cg20793071). This amplicon is located in the promoter of the gene that most statistically distinguished LOAD cases and controls on the Illumina array. Primers were complementary to bisulfite treated DNA in regions without CpG nucleotides (Table 3.2). The region of interest was amplified by bisulfite-PCR with the following thermal cycling protocol: 15 minute activation at 95ºC, 45 cycles of 30 second denaturation at 94ºC, 30 second annealing at 55ºC, and 1 minute extension at 72 minutes, followed by a final extension for 10 minutes at 72ºC. Serial dilutions of 100% methylated and unmethylated controls were used to test for any bias in amplification for each assay. Internal bisulfite conversion quality controls were incorporated at original sequence non-CpG cytosines by including C nucleotides in the dispensation order, which should be fully converted to T’s following bisulfite treatment.

Incorporation of either a T (for an unmethylated cytosine) or C (for a methylated cytosine) at each CpG provides a quantitative measure for consecutive CpG sites throughout the region sequenced. The level of methylation for each CpG within the target region of analysis was quantified using the Pyro Q-CpG Software. Primers and pyrosequencing assay file information including nucleotide dispensation orders and sequences to analyze are in Table 3.2.

Gene Expression

Functional relevance of top methylation marks distinguishing Alzheimer’s disease cases and controls was assessed via SYBR green Real Time PCR gene expression assays. RNA was extracted from all 25 matched pair samples using an adjacent portion of the same gray matter sample used for DNA. RNA was extracted using the Qiagen RNeasy Lipid Tissue Kit (Valencia, CA), following homogenization with the Qiagen Tissue Lyser instrument. Assays were designed using Genscript software (Piscataway, NJ). cDNA was generated with the Bio Rad iScript cDNA Synthesis Kit (Hercules, CA) and the primers are listed

in Table 3.3. Quantitative PCR assays were run with the iQ SYBR Green Supermix (Bio Rad) on the CFX96 C1000 Thermal Cycler (Bio Rad). CFX

Manager software (Bio Rad) was used to determine the threshold cycle (CT) and perform inter-plate normalizations. CT values relative to β-actin levels were used to compute a fold change between matched pairs. A subset of samples was analyzed on the Affymetrix GeneChip Human Genome U133 Plus 2.0 Array in the University of Michigan Sequencing Core using standard protocols. We determined that β-actin is a suitable control in these samples as gene expression of β-actin did not differ by case control status at any of the 6 probe sets included in the Affymetrix Array (p-values: 0.13, 0.18, 0.23, 0.43, 0.48, 0.77). Gene expression data were evaluated for normality using R Statistical Software. To determine if higher methylation values were associated with decreased expression of target genes, Spearman correlation coefficients were calculated between CpG methylation (as measured by pyrosequencing) and the expression level data.

Protein Quantification

Adjacent portions of the same 25 matched-pair, gray matter tissue used for DNA were homogenized and protein extracted in Thermo Scientific RIPA buffer (Burlington, Ontario). Protein concentration was quantified using the Thermo Scientific Pierce BSA assay (Burlington, Ontario). Protein (25 ug) was loaded on 10% SDS-polyacrylamide gels for Western Blot analysis. Transmembrane protein 59 antibody was purchased from Novus Biologicals (Littleton, CO) and Transmembrane protein 59 control protein was purchased from OriGene (Rockville, MD). The blots were imaged on the VersaDoc 5000MP instrument (Bio Rad) with Quantity One densitometry software (Bio Rad) under a consistent, constrained area. The levels of Transmembrane protein 59 were standardized to the corresponding tubulin band.

RESULTS

Genome-Wide Descriptive Statistics

DNA was extracted from the frontal cortex of 25 age- and gender-matched LOAD case and control human post-mortem pairs (Table 3.1). Sixteen pairs were male and nine pairs were female. LOAD cases and controls did not differ by age (p-value = 0.68). The mean post-mortem interval (PMI) was 12.8 hours, ranging from 3-28 hours. Controls had significantly longer PMI than LOAD cases (p-value=0.0004). The mean number of years in storage was 11.8, ranging from 2-21 years. Cases and controls did not statistically differ in the number of years in storage (p=0.07).

In the Discovery Set, a randomly selected subset of 12 matched pairs ranging in age from 69-95 (mean age 79.8) were analyzed for genome-wide DNA methylation using the Illumina Infinium HumanMethylation27 BeadArray. The BeadArray represented 27,578 CpG sites corresponding to 14,475 unique genes. The average number of CpG sites per gene was 1.9, and 92.0% of CpG sites were within 1000 bp of a transcription start site. CpG sites within CpG islands were overrepresented on the array, as 72.5% of sites were within CpG islands. The average number of CpG sites per sample with Illumina detection p-values greater than 0.05 (considered failing) was 71.1 (range 11-360 sites). CpG sites that failed on more than 2 samples (>10% samples) were not included for further analyses (n= 171 sites).

The global distribution of 5’-cytosine modifications at all CpG sites measured by the array was bimodal, and the distribution of methylation levels was very similar between cognitively normal controls (Figure 3.1A) and Alzheimer’s disease cases (Figure 3.1B). There was a large cluster of sites that had less than 10% methylation (15,735 in controls and 15,619 in AD cases) based on the mean of each group of 12 samples. A second cluster of sites was modified between 75% and 100% (5,226 sites in controls and 5,162 sites in AD cases).

Of the 27,578 sites on the array, 25,380 sites were located in promoter regions as defined within 1,000 bp of a transcription start site. Only 2,198 sites were outside of known promoters. Promoter CpG sites had a median of 5.1%

methylation (IQR= 2.1-5.2) while non-promoter sites had a median of 59.7% methylation (IQR=11.0-84.0). This observation did not depend on AD case status. On the array, 20,006 sites were located within CpG islands and 7,572 sites were outside of CpG islands. CpG island sites had a median methylation of 3.2% (IQR = 1.7-9.6) and non-CpG island sites had 74.6% median methylation (IQR=45.5-85.4). This did not depend on AD case status. Our discovery set of 24 samples ranged in age from 69-95 years (Table 3.1), and age was an important predictor of methylation. There were 2,416 sites and 2,227 unique genes associated with age (based on a p-value of 0.05). Of these sites, 1,294 were hypermethylated with increasing age and 1,122 were hypomethylated with age. The top 25 CpG sites associated with age among controls are listed in Table 3.4.

Alzheimer’s Disease-Specific Results

Following age and gender adjusted linear fit models with empirical Bayseian standard error adjustments, 948 CpG sites representing 918 unique genes were significantly associated with AD case status (based on p-value of 0.05). Among these 948 sites, the maximum mean methylation difference between AD cases and controls was 19.5% at a CpG site 249 base pairs from the predicted TSS of C21orf56 on chromosome 21 (cases 34.8% methylated vs. controls 15.9% methylated). The mean observed disease specific methylation difference across the 958 sites was 2.9% (IQR=0.88-4.2).

The top 26 autosomal CpG sites distinguishing cases and controls (as defined by FDR) are depicted in a heatmap (Figure 3.2). Samples clustered on case status with the exception of two control samples. One of those controls was the oldest control subject in the study at 94 years of age. The top 25 CpG sites that were significantly different by case status are outlined in Table 3.5.

Gene ontology analysis of the top 958 disease specific sites revealed hypermethylation in several molecular functions and biological processes associated with transcription. The top 10 molecular functions enriched for

hypermethylation with AD cases were RNA polymerase II transcription factor activity (Figure 3.3A), RNA binding, GTPase regulator activity, cytokine binding, DNA binding, lyase activity, ATPase activity, transcription factor activity, and nucleoside triphosphatase activity. Similarly, the top 10 biological processes associated with hypermethylation were nucleic acid metabolic process, DNA replication, regulation of nucleic acid metabolism, regulation of DNA dependent transcription, regulation of RNA metabolic process, regulation of cell cycle, DNA dependent transcription, positive regulation of RNA metabolic process, DNA metabolic process, RNA biosynthetic process, and nervous system development.

Hypomethylation was enriched at functions and processes related to membrane transport and protein metabolism. The top 10 molecular functions associated with hypomethylation in AD cases are electron carrier activity, cation transmembrane transporter activity, metal ion transmembrane transporter activity, enzyme binding, rhodopsin-like receptor activity, cation channel activity, integrin binding, phosphoric ester hydrolase activity, G-protein coupled receptor activity, and peptidase activity. The top 10 biological processes associated with hypomethylation in LOAD cases were carboxylic acid metabolic processes (Figure 3.3B), organic acid metabolic process, biosynthetic process, cation transport, nitrogen compound metabolic process, amine metabolic process, negative regulation of developmental process, programmed cell death, apoptosis, and anti-apoptosis.

Several promoter and 3’ UTR regulatory binding motifs were enriched in the disease associated gene list. Hypermethylation in LOAD cases was observed in genes containing binding site motifs for transcription factors POU3F2 (p-value < 0.001) and HOXA4 (p-value=0.004), and microRNAs MIR-9 (p-value = 0.002), MIR-518C (p-value < 0.001), MIR-1 (p-value=0.025), and MIR-326 (p- value=0.019). Genes containing MIR-140 (p-value = 0.04) and NFE2 (p- value=0.019) motifs were hypomethylated in LOAD cases.

Positional gene set analysis of the 958 disease associated CpG sites identified enrichment of hypomethylation at the chromosomal location19q13

(Normalized Enrichment Score (NES) = 1.24), in cytogenetic band region where the APOE gene is located (Kim et al. 2009). In addition, hypermethylation in LOAD cases was seen at 19q13 (NES= -0.72), a candidate location for genetic linkage with LOAD (Wijsman et al. 2004). We visually identified broad regions of altered methylation in AD brains compared to control brains (Figure 3.4). These include the p-arm of chromosome 14 and distal q-arm of chromosome 3 (hypomethylated in AD brains compared to control brains) and the p-arms of chromosomes 10 and 17 (hypermethylated in AD brains compared to control brains). Chromosome 15 had the highest density of observed disease-specific methylation differences.

We checked the list of disease specific hits identified with the BeadArray for genes known to be associated with AD. There were two CpG sites on the array corresponding to each of the following genes: ABCA7, APOE, AΒPP, BACE1, BDNF, BIN1, CD2AP, CD33, CLU, CR1, EPHA1, MAPT, MS4A6A, PICALM, PSEN2, and TOMM40. There was one site representing PSEN1 on the array and this site was associated with AD (Cases mean methylation = 1.6%, Controls mean methylation = 2.6%; p-value=0.034; cg11490446). Gene expression of the probe set for PSEN1 at exon 2 differed by case status based on the results of the Affymetrix gene expression array (probe 207782_s_at, p- value = 0.0076, fdr = 0.35). The Spearman correlation coefficient linking expression of this gene expression probe set and methylation measured by the Illumina BeadArray is -0.61 (pvalue=0.0014). AD cases were less methylated and had higher expression of this probe. There was no difference gene expression in the other five probe sets for PSEN1. One of the two sites corresponding to EPHA1 was associated with hypermethylation with age (p-value = 0.029; cg02376703). One of the two sites associated with PSEN2 was associated with hypomethylation with age (p-value = 0.030; cg25514304). The remaining CpG sites within LOAD candidate genes were not associated with differential methylation by case status or age.

We also assessed known human imprinted genes for their association with AD, focusing on CpG sites located within Differentially Methylated Regions (DMRs) (Choufani et al. 2011). The Illumina array contains 10 sites in the DMR for DIRAS3, 8 sites each for PLAGL1 and GNAS, 7 sites for ZIM2, 4 sites for PEG10, 3 sites for PEG3, 2 sites for MEST, and 1 site each for the genes GRB10, KCNQ1, and SNURF. One of the sites in the DMR for DIRAS3 was more highly methylated in AD cases (43.4%) than controls (38.5%) (p-value = 0.024; cg21808053). One of the sites in the DMR for GNAS was hypomethylated with age among controls (p-value = 0.012; cg21625881) and the site for KCNQ1 was hypermethylated with age (p-value = 0.023; cg27119222).

Gene-Specific Results

After adjusting for age and sex, the highest ranking site (FDR = 0.36, p- value = 0.000013) associated with LOAD was a CpG site upstream of Transmembrane Protein 59 (TMEM59). TMEM59 is responsible for post- translational glycosylation of AΒPP and leads to retention of AΒPP in the Golgi apparatus (Ullrich et al. 2010). AD cases had 7.3% lower methylation at TMEM59 than controls, and this difference was more profound in older subjects (Figure 3.5A). Methylation of TMEM59 was significantly associated with age in cases relative to controls (p-value=0.013). In a second TMEM59 model, we added PMI as a predictor and compared the goodness of fit of the two nested models using an F-test. PMI did not improve the model fit and PMI is not a statistically significant predictor of methylation at the TMEM59 site. In a simple linear regression model with Case Status as the only predictor of TMEM59 methylation, the model fit R2 was 0.597. The model fit R2 with PMI as the only predictor was 0.132.

The methylation findings were technically validated by pyrosequencing a single assay containing both CpG sites from the Illumina array that were associated with TMEM59. The CpG density 1000 bp flanking the top hit on either side is 1.5%. Pyrosequencing the original 24 discovery samples confirmed the difference between cases and controls at TMEM59 was 2.7% (Figures 3.5B and

3.5D). Methylation was again associated with age in cases relative to controls in the technical validation (p-value=0.0084). Pyrosequencing of an additional 26 matched pair samples across an expanded younger age range (61-94) confirmed the age-associated reduction in methylation (p-value-0.0022), and the association with AD case status was not statistically significant at an alpha of 0.05 (p-value = 0.088) (Figure 3.5C).

We determined expression levels of TMEM59 at four points along the 8 exon transcript (Figure 3.6A) (including the beginning, end, and two middle sites) to functionally validate the DNA methylation results with the TMEM59 gene. At the four locations along the transcript that were assayed by real time PCR, controls had lower RNA expression levels than cases (Figure 3.6B). For the first assay on the transcript, AD cases had 24.9% higher expression (fold change 1.33; p-value 0.0013). For two assays in the middle of the transcript, AD cases had 20.5% (fold change 1.26; p-value 0.0071) and 28.3% (fold change 1.40; p- value 0.056) higher expression. The assay at the end of the transcript showed AD cases to have 21.5% (fold change 1.27; p-value-0.0036) higher expression than controls. DNA methylation and RNA expression were negatively correlated at TMEM59 (Spearman correlation coefficient = -0.274, p-value = 0.0083).

To further investigate the functional implications of the observed DNA methylation and gene expression differences of TMEM59, we measured the protein levels by Western blot in the full set of 25 case brains and 25 control brains. No differences were observed for the full length 36 kDa protein (p- value=0.68) (Figure 3.6C), but AD cases had reduced levels of a shorter protein that was also bound by the antibody specific for the C-terminus of TMEM59 (p- value=0.040) (Figures 3.6C, 3.6E). The quantity of shorter protein was associated with age (Figure 3.6D).

DISCUSSION

We performed a genome-wide, semi-unbiased quantitative comparison of frontal cortex DNA methylation from age- and gender-matched LOAD cases and

controls. The CpG site most strongly associated with case status was located in the promoter region of TMEM59, a gene recently implicated in Alzheimer’s disease pathogenesis (Ullrich et al. 2010). This gene is involved in the post- translational modification of Amyloid Precursor Protein (AβPP) and causes retention of AβPP in the Golgi apparatus (Ullrich et al. 2010). The magnitude of methylation difference at this site between cases and controls was very modest (7.3% difference using the Illumina array), but the direction of association was confirmed using an alternate method of DNA methylation detection (2.7% difference using pyrosequencing). In an expanded population including a higher number of younger cases and controls, an interaction between age and case status was detected. Thus, age modifies the association between disease status and methylation. There was not a primary association between case status and methylation when the younger population was included. In the original sample of LOAD cases and controls, TMEM59 DNA methylation levels corresponded to functional changes in TMEM59 gene expression. LOAD cases had lower methylation and higher expression of TMEM59 than control samples. No differences in the level of the full length TMEM59 protein were observed between cases and controls; however a smaller protein that bound theTMEM59 antibody was significantly higher in controls. This TMEM59 protein size pattern is consistent with the TMEM59 control protein lysate. The shorter protein may represent a proteolytic product of the full length protein. The shorter protein is approximately 17 kDa, which could also be consistent with translation of an alternative RNA transcript beginning at exon 5 of the TMEM59 gene. The observed differences in protein expression levels are consistent with epigenetic regulation. Further molecular research is needed to better understand the gene expression and protein regulation and potential role of DNA methylation at this site.

Well-studied genes known to be involved in AD pathogenesis or identified through GWAS for genetic association with LOAD were largely not associated with disease-specific DNA methylation differences in this study. A notable exception was PSEN1, which was modestly hypomethylated in LOAD cases.

Consistent with previous human post-mortem tissue studies, PSEN1 showed very low levels of methylation in our population (Siegmund et al. 2007). Here, LOAD cases had reduced DNA methylation that was associated with increased PSEN1 gene expression, suggesting the DNA methylation change may be functional at this site. Studies in mice and neuroblastoma cell lines demonstrate that PSEN1 gene expression is regulated by DNA methylation at specific promoter CpG sites and this regulation depends on B vitamin availability (Fuso et al. 2011). The correlation between DNA methylation and gene expression in our study support the cell line and mouse model findings. The Illumina HumanMethylation27 BeadArray platform used in this study also allowed for discovery of novel gene associations with AD. For example, methylation change was observed within the DMR for the imprinted gene, DIRAS3. Genomic imprinting in Alzheimer’s disease is a potential mechanism to explain epidemiological parent-of-origin inheritance observations(Fallin et al. 2011; Mosconi et al. 2011).

Greater than 900 genes (6% of genes featured on the array) were differentially methylated by case status after adjusting for age and gender. Overall, the disease related methylation effect size (2.9%) was relatively modest, and the global methylation distributions of AD cases and controls were similar. Together these findings suggest that DNA methylation may play a role in LOAD and the individual effects at each CpG site may be subtle. The magnitude and absolute number of DNA methylation changes observed in this study are consistent with previous reports in the literature performed on candidate gene subsets. In a case-control study of prefrontal cortex DNA methylation of twelve genes, only two genes were associated with AD status and the differences in methylation were less than 10% (Wang et al. 2008).

Gene set enrichment analysis revealed key patterns in the identified set of disease associated CpG sites. First, gene ontology analysis showed hypermethylation of genes involved in transcription and DNA replication, while membrane transporters were hypomethylated. Second, hypermethylation was

enriched at genes containing POU3F2 binding motifs. POU3F2 is a transcription factor critical in central nervous system development that regulates Nestin gene expression, a protein important for radial axon growth (Jin et al. 2009). Third, positional analysis showed hypomethylation with case status at 19q13 and hypermethylation at 19p13, cytogenic band regions genetically linked with AD (Kim et al. 2009; Wijsman et al. 2004).

LOAD cases and control sample groups were similar with respect length of time in storage, but LOAD cases had shorter PMI then control samples (p- value = 0.0004). During PMI, samples may be exposed to damaging lower pH conditions and higher temperatures where enzymes may be active. PMI was not a significant confounder at the TMEM59 site, however, PMI may be a factor at other specific locations throughout the DNA methylome.

This study measured genome-wide DNA methylation differences between LOAD case and control subjects aged 69 to 95. Across this relatively short age range, DNA methylation was associated with age at over 2,400 CpG sites, representing more than 8% of the sites on the BeadArray. Both hyper and hypomethylation was observed. Previous studies have observed global hypomethylation with age in the brain (Pogribny and Vanyushin 2010), but gene- specific studies of aging and methylation have noted varying patterns (Siegmund et al. 2007). These results further support age as an important covariate to consider in statistical models of DNA methylation in late life.

Age is a major factor in epigenetic change in the brain (Hernandez et al. 2011), potentially confounding or modifying disease specific associations. In a study of cerebral cortex DNA from gestation to 104 years of age, eight of fifty loci showed increases in methylation through late life and two sites presented changes suggestive of an acceleration of age-related change in a subset of samples with LOAD (Siegmund et al. 2007). Additional evidence supports increased age-dependent epigenetic drift with disease. In LOAD prefrontal cortex samples representing a 30 year age range, an age-specific epigenetic drift was more prominent among cases compared to controls. The average

methylation in promoters of MTHFR and APOE increased by 6.8%, while control samples decreased with age by 10.6% (Wang et al. 2008).

Cultured cells can potentially have very different epigenetic profiles than primary cells as an artifact of growth in culture (Smiraglia et al. 2001), and thus use of primary human frontal cortex tissue is a strength of this study. DNA methylation is brain region specific and greater differences have been seen between the cerebellum and cortex regions than by sex, age, post-mortem interval, race, or cause of death (Ladd-Acosta et al. 2007). This study consistently used frontal cortex tissue because of its role in advanced AD. As with many studies of epigenomic, transcriptomic, or proteomic changes in the human brain, the tissue samples represented populations of mixed cell types, an important limitation, which may have resulted in attenuated effects. Epigenomes are cell type specific (Lister et al. 2009; Maunakea et al. 2010), and brain cell types have different roles in AD (Selkoe 2001). The AD brain has an active changing cell population including neuronal loss and glial activation (Akiyama et al. 2000) that may in part be responsible for the observed results. DNA methylation data, however, was not enriched in inflammatory mediators, which would have supported changes due to gliosis. This study considered brain region specific methylation and as epigenomic platforms require lower input DNA amounts, future research may be able to also consider cell type specific changes.

Results from large DNA methylome and transcriptome maps of the human brain suggest that intragenic CpG sites rather than promoter CpG islands may better correlate with transcription (Maunakea et al. 2010). The genome-wide sites included on the Illumina Infinium HumanMethylation27 BeadArray are more likely to be located within promoter region CpG islands. Important methylation events located elsewhere throughout the genome would be missed using this platform and may be included in future research utilizing alternative technologies.

These results must be interpreted with caution because this study had a small sample size relative to the large number of CpG site comparisons and the

magnitude of observed methylation differences between LOAD cases and controls was moderate. The results from the array were technically validated at the top CpG site, but it is not clear whether this observation will be consistent across populations. While a small study, we identified modest DNA methylation differences as a potential event in LOAD.

FIGURES

Figure 3.1. Mean percent methylation frequency distribution of the Discovery Set of 12 cognitively normal control samples (A) and 12 Alzheimer’s disease cases (B) across the 27,578 CpG sites on the Illumina HumanMethylation27 BeadArray.

Figure 3.2. Hierarchical clustering heatmap of the Discovery Set top 26 autosomal CpG loci associated with late-onset Alzheimer’s disease (LOAD) case/control status after adjusting for sex and age. Green represents hypermethylation in LOAD cases and red represents hypomethylation in cases. Horizontal color bars at the top refer to the age, sex, and case status of the sample. In the Case Status color bar, light green represents control samples and dark green represents LOAD cases. For sex, female is light pink and male is dark blue. In the age panel, black represents ages 91-95, darkest gray 86-90, medium gray 81-85, light gray 76-80, lightest gray 71-75, and white represents ages 66-70. Vertical color bars on the left refer to the CpG island and promoter status of the CpG sites. In the CpG island bar, dark purple represents sites within CpG islands and light purple represents sites outside of CpG islands. In the promoter bar, dark orange represents sites within promoter regions and light orange represents sites outside of promoter regions.

Figure 3.3. Discovery Set gene set enrichment analysis plots. (A) Genes associated with RNA polymerase II transcription factor activity molecular function were hypermethylated in LOAD cases relative to controls (p-value = 0.013). (B) Genes associated with carboxylic acid metabolic biological processes were hypomethylated in LOAD cases relative to controls (p-value= 0.013).

Figure 3.4. Human chromosome ideogram in black. Distribution of CpG sites featured on the Illumina HumanMethylation27 BeadArray is below the chromosomes in blue. Distribution of CpG sites that were significantly associated with late-onset Alzheimer’s disease (LOAD) in the Discovery Set are above the ideograms. Green represents hypermethylation with LOAD status. Red represents hypomethylation with LOAD status.

100

Figure 3.5. Methylation upstream of the TMEM59 gene. (A) Percent methylation by age and case status (Late-Onset Alzheimer’s Disease cases in red; Controls in blue). Data from the Illumina HumanMethylation27 BeadArray. (B) Age vs. percent methylation bisulfite-pyrosequencing technical validation data of original 24 samples run in duplicate in the Discovery Set. (C) Age vs. percent methylation bisulfite pyrosequencing of Discovery Set 24 samples plus 26 additional population validation samples run in duplicate. (D) A representative bisulfite-pyrosequencing assay pyrogram for 2 CpG sites in the promoter of TMEM59.

101

Figure 3.6. Functional validation of observed DNA methylation differences for TMEM59, a gene involved in the post-translational modification of Amyloid Precursor Protein (AΒPP). (A) TMEM59 is located on chromosome 1 and is transcribed on the reverse strand. The reference sequence mRNA is yellow.

102

Predicted alternative isoforms are in blue. (B) Boxplot of TMEM59 gene expression by Q-PCR in the Discovery Set. Two-sample t-test between cases and controls were all statistically significant (exon 1 p-value=0.0013; exons 1-2 p- value=0.0071; exons 3-4 p-value=0.0036, exons 7-8 p-value=0.0083) (C) Boxplot of relative protein levels of TMEM59 in the Discovery Set plus an additional 26 validation samples. Paired t-tests did not reflect case specific differences for the full length protein (p-value=0.68), while the shorter protein fragment was significantly lower in AD cases (p-value=0.040). (D) Levels of the shorter TMEM59 protein fragment as a function of age. (E) Representative western blot image of TMEM59 protein expression in controls and AD cases 1-3 representing identical exposures of the same gel. No differences were detected between AD and controls for full length TMEM59 protein based on case status, but the levels of the TMEM59 shorter proteins were reduced in AD cases. These shorter proteins were also observed in the TMEM59 control protein lysate.

103

TABLES

Table 3.1. Study population mean demographics by case status. Range is provided in parentheses. LOAD Cases Cognitively Normal Controls Characteristic Discovery Set Full Set Discovery Full Set (n=12) (n=25) Set (n=12) (n=25) Neuropathological High High Control n=11 Control n=24 Diagnosis Likelihood AD Likelihood Other n=1 Other n=1 n=11 AD n=21 Intermediate Intermediate Likelihood Likelihood n=1 n=4 Braak Stage 4.7 (2-6) 4.8 (2-6) 1.3 (1-2) 1.5 (1-2) Age 79.6 (69-94) 78.2 (61-94) 79.9 (69-95) 78.3 ( 61-95) Sex Female 6 9 6 9 Male 6 16 6 16 Post Mortem 9.6 (3-24) 9.5 (3-27.5) 16.3 (6-24.5) 16.0 (5-28) Interval (hours) Years in Storage 10.75 (5-17) 10.5 (3-18) 13.2 (3-20) 13.2 (2-21) Race Caucasian Caucasian Caucasian Caucasian (n=12) (n=25) (n=12) (n=25) Age of onset 69.1 (59-78) NA NA MMSE 11.7 (0-28) 12.9 (0-30) 26.8 (25-28) 26 (24-28) missing=1 missing=2 missing=10 missing=16 # Years from 5.1 (1-12) 5.2 (1-12) NA NA Diagnosis to Death APOE Genotype 2/2 2 2 3 6 2/3 1 1 0 1 2/4 0 1 0 0 3/3 3 7 6 13 3/4 5 13 1 3 4/4 1 1 1 1 Missing 0 0 1 1

104

Table 3.2. Pyrosequencing assay information. Gene Sequence 5 – 3’ TMEM59 Forward Primer GGGTAGGTATATAGAATTATATTTGGTATT Reverse Primer AAATTTCTACACACCCCTACTACA Biotinylated Sequencing Primer AATAGATTATATTTTGTAAAAGAA Dispensation Order ATATCGATCGAGGATGTTGATCGAG Sequence to Analyze TAATAAYGAAGGGGATTTGTTTTAYGAGTTAGTATATATGGTGTAAAT

Table 3.3. Primer sequences for gene expression QPCR assays. Gene Sequence 5’ – 3’ TMEM59 Exon 1 Forward Primer TGACTCGGTCTTGGGTGATA Reverse Primer TCTTCCTTAGGGTAGGTGTGC TMEM59 Exons 1-2 Forward Primer GGGCCTGTCAGTTGACCTAC Reverse Primer CTGCAACCTCTCTGACATGC TMEM59 Exons 3-4 Forward Primer GAACAACTTATGTCCCTGATGC Reverse Primer CGTCATCGGCTTGAAGATAA TMEM59 Exons 7-8 Forward Primer TCCTCTCGGTGATGGTATTG Reverse Primer TCAGCTTCTCAGAGGGAACA Β-actin Forward Primer TGCTATCCAGGCTGTGCTAT Reverse Primer AGTCCATCACGATGCCAGT

105

Table 3.4. CpG sites differentially methylated with age among cognitively normal controls (Discovery Set, Age 69-94).Beta coefficient can be interpreted as the rate of change in methylation per year across the years studied. Distance to Associated Beta CpG Rank p-value Transcription Chromosome Biological Description Gene Coefficient Island Start Site Dynein intermediate polypeptide 2; axonemal;. Function in microtubule motor 1 DNAI2 0.441288 1.08E-05 714 FALSE 17 activity. Extracellular matrix protein 2 precursor. Function in integrin binding and cell- 2 ECM2 0.453176 3.85E-05 717 FALSE 9 matrix adhesion 3 UNQ689 0.409188 1.39E-04 991 FALSE 4 hypothetical protein LOC401138 Complement component 3 precursor. Function in acylation-stimulating protein 4 C3 0.262665 1.52E-04 680 FALSE 19 cleavage in innate immune response oncostatin M receptor. Role in cell proliferation and cell surface linked signal 5 OSMR 0.25763 2.10E-04 404 TRUE 5 transduction 6 MEG3 -0.45638 2.15E-04 NA TRUE 14 Predicted gene from GNOMON glyoxalase I. Role in zinc ion binding, lyase activity in carbohydrate metabolism 7 GLO1 0.238369 2.36E-04 480 TRUE 6 and antiapoptosis. 8 CRNN 0.295641 2.50E-04 167 FALSE 1 Hypothetical protein LOC49860. Tumor-related protein. 9 SF3B2 -0.16749 2.92E-04 484 TRUE 11 splicing factor 3B subunit 2 106 phosphoinositide-3-kinase; class 2; beta polypeptide. Role in intracellular 10 PIK3C2B 0.221434 3.22E-04 164 FALSE 1 signaling cascade 11 RIBC2 0.226036 3.68E-04 126 TRUE 22 RIB43A domain with coiled-coils 2. Synonym C22orf11 12 CCDC74B 0.230644 3.71E-04 1015 FALSE 2 hypothetical protein LOC91409 13 C20orf77 0.622206 6.42E-04 605 TRUE 20 hypothetical protein LOC58490 14 C9orf112 0.530497 6.84E-04 317 TRUE 9 hypothetical protein LOC92715 15 LCE1B 0.388947 7.51E-04 1310 FALSE 1 late cornified envelope 1B Role in epidermal differentiation complex 2A 16 SFRS11 0.36302 7.59E-04 1480 FALSE 1 Splicing factor p54. Nucleic acid binding and nuclear mRNA splicing solute carrier family 18 (vesicular monoamine); member 2. Vesicle monoamine 17 SLC18A2 0.055023 8.92E-04 275 TRUE 10 transporter type 2. pregnancy-associated plasma protein A preproprotein. Insulin-like growth 18 PAPPA -0.21564 1.07E-03 204 FALSE 9 factor dependent IGF binding protein. 19 FIGNL1 0.64861 1.08E-03 599 TRUE 7 fidgetin-like 1 ATP binding nucleoside-triphosphatase activity 20 NMT1 -0.08842 1.10E-03 285 TRUE 17 N-myristoyltransferase 1 vesicle-associated membrane protein 5 (myobrevin). Role in vesicle-mediated 21 VAMP5 -0.28073 1.13E-03 492 TRUE 2 transport, myogenesis, and cell differentiation 22 FLJ33641 0.634 1.20E-03 974 FALSE 5 hypothetical protein LOC202309 23 DVL3 0.40801 1.21E-03 580 TRUE 3 dishevelled 3. Kinase activity. Role in nervous system development. 24 C20orf4 -0.15153 1.22E-03 250 TRUE 20 hypothetical protein LOC25980 25 IGF2 0.534292 1.25E-03 NA TRUE 11 insulin-like growth factor 2

Table 3.5. Table of the 25 CpG sites most significantly differentially methylated by AD case status (Discovery Set). Distance to % Methylation % Methylation Chromoso Rank Associated Gene p-value Transcription CpG Island Biological Description Cases Controls me Start Site 1 TMEM59 63.03 70.30 1.32E-05 1339 FALSE 1 APP post-translational glycolytic processing 2 ATG10 8.16 5.59 1.97E-04 197 TRUE 5 Autophagy 3 C9orf138 85.53 77.88 2.08E-04 47 TRUE 9 Hypothetical Protein 4 CPNE9 4.76 6.33 4.43E-04 549 TRUE 3 Copine-like protein reticuloendotheliosis viral oncogene homolog 5 RELB 36.40 45.75 5.68E-04 470 TRUE 19 B 6 C9orf138 68.36 57.60 8.92E-04 406 TRUE 9 Hypothetical Protein 7 PLA2G3 45.95 37.03 9.66E-04 488 FALSE 22 Phospholipase A2 8 DHFRL1 6.41 4.41 1.08E-03 511 TRUE 3 Hypothetical Protein 9 MBD3L1 16.33 11.98 1.08E-03 141 FALSE 19 Methyl-CpG binding domain protein 3-like Restin isoform a. Intermediate filament 10 RSN 2.37 2.85 1.37E-03 698 TRUE 12 associated protein 11 OTUD5 22.39 18.53 1.53E-03 232 TRUE X Hypothetical Protein Tubulin, beta polypeptide paralog. Microtubule 12 TUBB2B 7.58 10.50 1.53E-03 494 TRUE 6 associated.

107 Netrin 2-like. Structural molecule, axon 13 NTN2L 7.63 4.97 1.59E-03 159 TRUE 16 guidance.

14 GPR142 90.60 88.52 1.64E-03 237 FALSE 17 Signal transduction 15 TSCOT 55.93 62.95 1.65E-03 498 TRUE 9 Thymic stromal co-transporter 16 IL2RG 67.12 60.80 1.70E-03 88 FALSE X Interleukin 2 receptor, gamma precursor. Zinc finger protein basonuclin. Metal ion 17 BNC1 39.63 48.20 1.86E-03 NA TRUE 15 binding. 18 HERC5 2.57 1.43 1.91E-03 108 TRUE 4 Cyclin-E binding protein 1. 19 SLC36A3 73.03 80.25 2.18E-03 203 FALSE 5 Proton/amino acid transporter 3 20 DYNC2LI1 4.44 3.05 2.21E-03 24 TRUE 2 Dynein 2 light intermediate chain 21 SLC7A3 10.20 6.88 2.32E-03 208 TRUE X Cationic amino acid transporter 22 FGF5 4.24 5.70 2.36E-03 544 TRUE 4 Fibroblast growth factor 5 23 CAMP 92.62 89.57 2.53E-03 284 FALSE 3 Cathelicidin antimicrobial peptide 24 CNN1 13.81 9.18 2.59E-03 92 TRUE 19 Caloonin 1. Calmodulin binding 25 C15orf21 87.01 88.96 2.76E-03 7 FALSE 15 Dresden prostate carcinoma 2

REFERENCES

Akiyama H, Barger S, Barnum S, Bradt B, Bauer J, Cole GM, Cooper NR, Eikelenboom P, Emmerling M, Fiebich BL, Finch CE, Frautschy S, Griffin WS, Hampel H, Hull M, Landreth G, Lue L, Mrak R, Mackenzie IR, McGeer PL, O'Banion MK, Pachter J, Pasinetti G, Plata-Salaman C, Rogers J, Rydel R, Shen Y, Streit W, Strohmeyer R, Tooyoma I, Van Muiswinkel FL, Veerhuis R, Walker D, Webster S, Wegrzyniak B, Wenk G, Wyss-Coray T. 2000. Inflammation and Alzheimer's disease. Neurobiol Aging 21(3): 383-421. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, Harris MA, Hill DP, Issel-Tarver L, Kasarskis A, Lewis S, Matese JC, Richardson JE, Ringwald M, Rubin GM, Sherlock G. 2000. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 25(1): 25-29. Ashford JW. 2004. APOE genotype effects on Alzheimer's disease onset and epidemiology. J Mol Neurosci 23(3): 157-165. Barrachina M, Ferrer I. 2009. DNA methylation of Alzheimer disease and tauopathy-related genes in postmortem brain. J Neuropathol Exp Neurol 68(8): 880-891. Bertram L. 2009. Alzheimer's disease genetics current status and future perspectives. Int Rev Neurobiol 84: 167-184. Bibikova M, Le J, Barnes B, Saedinia-Melnyk S, Zhou LX, Shen R, Gunderson KL. 2009. Genome-wide DNA methylation profiling using Infinium (R) assay. Epigenomics-Uk 1(1): 177-200. Bird TD. 2005. Genetic factors in Alzheimer's disease. N Engl J Med 352(9): 862- 864. Choufani S, Shapiro JS, Susiarjo M, Butcher DT, Grafodatskaya D, Lou Y, Ferreira JC, Pinto D, Scherer SW, Shaffer LG, Coullin P, Caniggia I, Beyene J, Slim R, Bartolomei MS, Weksberg R. 2011. A novel approach identifies new differentially methylated regions (DMRs) associated with imprinted genes. Genome Res 21(3): 465-476. Chouliaras L, Rutten BP, Kenis G, Peerbooms O, Visser PJ, Verhey F, van Os J, Steinbusch HW, van den Hove DL. 2010. Epigenetic regulation in the pathophysiology of Alzheimer's disease. Prog Neurobiol 90(4): 498-510. Christensen H, Batterham PJ, Mackinnon AJ, Jorm AF, Mack HA, Mather KA, Anstey KJ, Sachdev PS, Easteal S. 2008. The association of APOE genotype and cognitive decline in interaction with risk factors in a 65-69 year old community sample. BMC Geriatr 8: 14.

108

Dolinoy DC, Jirtle RL. 2008. Environmental epigenomics in human health and disease. Environ Mol Mutagen 49(1): 4-8. Ertekin-Taner N. 2010. Genetics of Alzheimer disease in the pre- and post- GWAS era. Alzheimers Res Ther 2(1): 3. Fallin MD, Szymanski M, Wang R, Gherman A, Bassett SS, Avramopoulos D. 2011. Fine mapping of the chromosome 10q11-q21 linkage region in Alzheimer's disease cases and controls. Neurogenetics 11(3): 335-348. Fuso A, Nicolia V, Pasqualato A, Fiorenza MT, Cavallaro RA, Scarpa S. 2011. Changes in Presenilin 1 gene methylation pattern in diet-induced B vitamin deficiency. Neurobiol Aging 32(2): 187-199. Gatz M, Pedersen NL, Berg S, Johansson B, Johansson K, Mortimer JA, Posner SF, Viitanen M, Winblad B, Ahlbom A. 1997. Heritability for Alzheimer's disease: the study of dementia in Swedish twins. J Gerontol A Biol Sci Med Sci 52(2): M117-125. Hardy J. 1997. Amyloid, the presenilins and Alzheimer's disease. Trends Neurosci 20(4): 154-159. Harold D, Abraham R, Hollingworth P, Sims R, Gerrish A, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Williams A, Jones N, Thomas C, Stretton A, Morgan AR, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC, Gill M, Lawlor B, Lynch A, Morgan K, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Love S, Kehoe PG, Hardy J, Mead S, Fox N, Rossor M, Collinge J, Maier W, Jessen F, Schurmann B, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Hull M, Rujescu D, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Carrasquillo MM, Pankratz VS, Younkin SG, Holmans PA, O'Donovan M, Owen MJ, Williams J. 2009. Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer's disease. Nat Genet 41(10): 1088-1093. Hernandez DG, Nalls MA, Gibbs JR, Arepalli S, van der Brug M, Chong S, Moore M, Longo DL, Cookson MR, Traynor BJ, Singleton AB. 2011. Distinct DNA methylation changes highly correlated with chronological age in the human brain. Hum Mol Genet 20(6): 1164-1172. Hollingworth P, Harold D, Sims R, Gerrish A, Lambert JC, Carrasquillo MM, Abraham R, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Jones N, Stretton A, Thomas C, Richards A, Ivanov D, Widdowson C, Chapman J, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC,

109

Gill M, Lawlor B, Lynch A, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Beaumont H, Warden D, Wilcock G, Love S, Kehoe PG, Hooper NM, Vardy ER, Hardy J, Mead S, Fox NC, Rossor M, Collinge J, Maier W, Jessen F, Ruther E, Schurmann B, Heun R, Kolsch H, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Gallacher J, Hull M, Rujescu D, Giegling I, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Pankratz VS, Sando SB, Aasly JO, Barcikowska M, Wszolek ZK, Dickson DW, Graff-Radford NR, Petersen RC, van Duijn CM, Breteler MM, Ikram MA, Destefano AL, Fitzpatrick AL, Lopez O, Launer LJ, Seshadri S, Berr C, Campion D, Epelbaum J, Dartigues JF, Tzourio C, Alperovitch A, Lathrop M, Feulner TM, Friedrich P, Riehle C, Krawczak M, Schreiber S, Mayhaus M, Nicolhaus S, Wagenpfeil S, Steinberg S, Stefansson H, Stefansson K, Snaedal J, Bjornsson S, Jonsson PV, Chouraki V, Genier-Boley B, Hiltunen M, Soininen H, Combarros O, Zelenika D, Delepine M, Bullido MJ, Pasquier F, Mateo I, Frank-Garcia A, Porcellini E, Hanon O, Coto E, Alvarez V, Bosco P, Siciliano G, Mancuso M, Panza F, Solfrizzi V, Nacmias B, Sorbi S, Bossu P, Piccardi P, Arosio B, Annoni G, Seripa D, Pilotto A, Scarpini E, Galimberti D, Brice A, Hannequin D, Licastro F, Jones L, Holmans PA, Jonsson T, Riemenschneider M, Morgan K, Younkin SG, Owen MJ, O'Donovan M, Amouyel P, Williams J. 2011. Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer's disease. Nat Genet. Hong MG, Myers AJ, Magnusson PK, Prince JA. 2008. Transcriptome-wide assessment of human brain and lymphocyte senescence. PLoS One 3(8): e3024. International AsD. 2010. World Alzheimer Report 2010: The global economic impact of dementia. Jin SG, Wu X, Li AX, Pfeifer GP. 2011. Genomic mapping of 5- hydroxymethylcytosine in the human brain. Nucleic Acids Res. Jin Z, Liu L, Bian W, Chen Y, Xu G, Cheng L, Jing N. 2009. Different transcription factors regulate nestin gene expression during P19 cell neural differentiation and central nervous system development. J Biol Chem 284(12): 8160-8173. Kim J, Basak JM, Holtzman DM. 2009. The role of apolipoprotein E in Alzheimer's disease. Neuron 63(3): 287-303.

110

Ladd-Acosta C, Pevsner J, Sabunciyan S, Yolken RH, Webster MJ, Dinkins T, Callinan PA, Fan JB, Potash JB, Feinberg AP. 2007. DNA methylation signatures within the human brain. Am J Hum Genet 81(6): 1304-1315. Lambert JC, Heath S, Even G, Campion D, Sleegers K, Hiltunen M, Combarros O, Zelenika D, Bullido MJ, Tavernier B, Letenneur L, Bettens K, Berr C, Pasquier F, Fievet N, Barberger-Gateau P, Engelborghs S, De Deyn P, Mateo I, Franck A, Helisalmi S, Porcellini E, Hanon O, de Pancorbo MM, Lendon C, Dufouil C, Jaillard C, Leveillard T, Alvarez V, Bosco P, Mancuso M, Panza F, Nacmias B, Bossu P, Piccardi P, Annoni G, Seripa D, Galimberti D, Hannequin D, Licastro F, Soininen H, Ritchie K, Blanche H, Dartigues JF, Tzourio C, Gut I, Van Broeckhoven C, Alperovitch A, Lathrop M, Amouyel P. 2009. Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer's disease. Nat Genet 41(10): 1094-1099. Li YJ, Scott WK, Hedges DJ, Zhang F, Gaskell PC, Nance MA, Watts RL, Hubble JP, Koller WC, Pahwa R, Stern MB, Hiner BC, Jankovic J, Allen FA, Jr., Goetz CG, Mastaglia F, Stajich JM, Gibson RA, Middleton LT, Saunders AM, Scott BL, Small GW, Nicodemus KK, Reed AD, Schmechel DE, Welsh-Bohmer KA, Conneally PM, Roses AD, Gilbert JR, Vance JM, Haines JL, Pericak-Vance MA. 2002. Age at onset in two common neurodegenerative diseases is genetically controlled. Am J Hum Genet 70(4): 985-993. Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, Nery JR, Lee L, Ye Z, Ngo QM, Edsall L, Antosiewicz-Bourget J, Stewart R, Ruotti V, Millar AH, Thomson JA, Ren B, Ecker JR. 2009. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462(7271): 315-322. Loring JF, Wen X, Lee JM, Seilhamer J, Somogyi R. 2001. A gene expression profile of Alzheimer's disease. DNA Cell Biol 20(11): 683-695. Mastroeni D, Grover A, Delvaux E, Whiteside C, Coleman PD, Rogers J. 2008. Epigenetic changes in Alzheimer's disease: Decrements in DNA methylation. Neurobiol Aging. Mastroeni D, Grover A, Delvaux E, Whiteside C, Coleman PD, Rogers J. 2011. Epigenetic mechanisms in Alzheimer's disease. Neurobiol Aging. Mastroeni D, McKee A, Grover A, Rogers J, Coleman PD. 2009. Epigenetic differences in cortical neurons from a pair of monozygotic twins discordant for Alzheimer's disease. PLoS One 4(8): e6617. Maunakea AK, Nagarajan RP, Bilenky M, Ballinger TJ, D'Souza C, Fouse SD, Johnson BE, Hong C, Nielsen C, Zhao Y, Turecki G, Delaney A, Varhol R, Thiessen N, Shchors K, Heine VM, Rowitch DH, Xing X, Fiore C, Schillebeeckx M, Jones SJ, Haussler D, Marra MA, Hirst M, Wang T,

111

Costello JF. 2010. Conserved role of intragenic DNA methylation in regulating alternative promoters. Nature 466(7303): 253-257. Mill J. 2011. Toward an integrated genetic and epigenetic approach to Alzheimer's disease. Neurobiol Aging. Mill J, Tang T, Kaminsky Z, Khare T, Yazdanpanah S, Bouchard L, Jia P, Assadzadeh A, Flanagan J, Schumacher A, Wang SC, Petronis A. 2008. Epigenomic profiling reveals DNA-methylation changes associated with major psychosis. Am J Hum Genet 82(3): 696-711. Miller JA, Oldham MC, Geschwind DH. 2008. A systems level analysis of transcriptional changes in Alzheimer's disease and normal aging. J Neurosci 28(6): 1410-1420. Mosconi L, Tsui W, Murray J, McHugh P, Li Y, Williams S, Pirraglia E, Glodzik L, De Santi S, Vallabhajosula S, de Leon MJ. 2011. Maternal age affects brain metabolism in adult children of mothers affected by Alzheimer's disease. Neurobiol Aging. Naj AC, Jun G, Beecham GW, Wang LS, Vardarajan BN, Buros J, Gallins PJ, Buxbaum JD, Jarvik GP, Crane PK, Larson EB, Bird TD, Boeve BF, Graff- Radford NR, De Jager PL, Evans D, Schneider JA, Carrasquillo MM, Ertekin-Taner N, Younkin SG, Cruchaga C, Kauwe JS, Nowotny P, Kramer P, Hardy J, Huentelman MJ, Myers AJ, Barmada MM, Demirci FY, Baldwin CT, Green RC, Rogaeva E, George-Hyslop PS, Arnold SE, Barber R, Beach T, Bigio EH, Bowen JD, Boxer A, Burke JR, Cairns NJ, Carlson CS, Carney RM, Carroll SL, Chui HC, Clark DG, Corneveaux J, Cotman CW, Cummings JL, Decarli C, Dekosky ST, Diaz-Arrastia R, Dick M, Dickson DW, Ellis WG, Faber KM, Fallon KB, Farlow MR, Ferris S, Frosch MP, Galasko DR, Ganguli M, Gearing M, Geschwind DH, Ghetti B, Gilbert JR, Gilman S, Giordani B, Glass JD, Growdon JH, Hamilton RL, Harrell LE, Head E, Honig LS, Hulette CM, Hyman BT, Jicha GA, Jin LW, Johnson N, Karlawish J, Karydas A, Kaye JA, Kim R, Koo EH, Kowall NW, Lah JJ, Levey AI, Lieberman AP, Lopez OL, Mack WJ, Marson DC, Martiniuk F, Mash DC, Masliah E, McCormick WC, McCurry SM, McDavid AN, McKee AC, Mesulam M, Miller BL, Miller CA, Miller JW, Parisi JE, Perl DP, Peskind E, Petersen RC, Poon WW, Quinn JF, Rajbhandary RA, Raskind M, Reisberg B, Ringman JM, Roberson ED, Rosenberg RN, Sano M, Schneider LS, Seeley W, Shelanski ML, Slifer MA, Smith CD, Sonnen JA, Spina S, Stern RA, Tanzi RE, Trojanowski JQ, Troncoso JC, Van Deerlin VM, Vinters HV, Vonsattel JP, Weintraub S, Welsh-Bohmer KA, Williamson J, Woltjer RL, Cantwell LB, Dombroski BA, Beekly D, Lunetta KL, Martin ER, Kamboh MI, Saykin AJ, Reiman EM, Bennett DA, Morris JC, Montine TJ, Goate AM, Blacker D, Tsuang DW, Hakonarson H, Kukull WA, Foroud TM, Haines JL, Mayeux R, Pericak-Vance MA, Farrer LA, Schellenberg GD. 2011. Common variants at MS4A4/MS4A6E,

112

CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer's disease. Nat Genet. Nee LE, Lippa CF. 1999. Alzheimer's disease in 22 twin pairs--13-year follow-up: hormonal, infectious and traumatic factors. Dement Geriatr Cogn Disord 10(2): 148-151. Pan S, Shi M, Jin J, Albin RL, Lieberman A, Gearing M, Lin B, Pan C, Yan X, Kashima DT, Zhang J. 2007. Proteomics identification of proteins in human cortex using multidimensional separations and MALDI tandem mass spectrometer. Mol Cell Proteomics 6(10): 1818-1823. Pogribny IP, Vanyushin BF. 2010. Age-Related Genomic Hypomethylation. Epigenetics of Aging: 11-27. Selkoe DJ. 2001. Alzheimer's disease: genes, proteins, and therapy. Physiol Rev 81(2): 741-766. Siegmund KD, Connor CM, Campan M, Long TI, Weisenberger DJ, Biniszkiewicz D, Jaenisch R, Laird PW, Akbarian S. 2007. DNA methylation in the human cerebral cortex is dynamically regulated throughout the life span and involves differentiated neurons. PLoS One 2(9): e895. Smiraglia DJ, Rush LJ, Fruhwald MC, Dai Z, Held WA, Costello JF, Lang JC, Eng C, Li B, Wright FA, Caligiuri MA, Plass C. 2001. Excessive CpG island hypermethylation in cancer cell lines versus primary human malignancies. Hum Mol Genet 10(13): 1413-1419. Smyth GK. 2004. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3: Article3. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP. 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102(43): 15545-15550. Suter MA, Aagaard-Tillery KM. 2009. Environmental influences on epigenetic profiles. Semin Reprod Med 27(5): 380-390. Tohgi H, Utsugisawa K, Nagane Y, Yoshimura M, Genda Y, Ukitsu M. 1999a. Reduction with age in methylcytosine in the promoter region -224 approximately -101 of the amyloid precursor protein gene in autopsy human cortex. Brain Res Mol Brain Res 70(2): 288-292. Tohgi H, Utsugisawa K, Nagane Y, Yoshimura M, Ukitsu M, Genda Y. 1999b. The methylation status of cytosines in a tau gene promoter region alters

113

with age to downregulate transcriptional activity in human cerebral cortex. Neurosci Lett 275(2): 89-92. Ullrich S, Munch A, Neumann S, Kremmer E, Tatzelt J, Lichtenthaler SF. 2010. The novel membrane protein TMEM59 modulates complex glycosylation, cell surface expression, and secretion of the amyloid precursor protein. J Biol Chem 285(27): 20664-20674. Wang SC, Oelze B, Schumacher A. 2008. Age-specific epigenetic drift in late- onset Alzheimer's disease. PLoS One 3(7): e2698. Wijsman EM, Daw EW, Yu CE, Payami H, Steinbart EJ, Nochlin D, Conlon EM, Bird TD, Schellenberg GD. 2004. Evidence for a novel late-onset Alzheimer disease locus on chromosome 19p13.2. Am J Hum Genet 75(3): 398-409. Xie X, Lu J, Kulbokas EJ, Golub TR, Mootha V, Lindblad-Toh K, Lander ES, Kellis M. 2005. Systematic discovery of regulatory motifs in human promoters and 3' UTRs by comparison of several mammals. Nature 434(7031): 338-345.

114

CHAPTER IV Research Chapter 3

An integrated analysis of genome-wide RNA expression and DNA methylation in late-onset Alzheimer’s disease and neuropathological controls

ABSTRACT

Introduction: Spontaneous, neurodegenerative, late-onset Alzheimer’s disease (LOAD) is associated with aberrations in brain gene expression. Recent studies have observed dysregulation of epigenetic mechanisms, such as DNA methylation, but the association of gene expression with these marks in disease tissue is unknown. Gene expression analysis in conjunction with DNA methylation mapping provides insight into the mechanisms of LOAD.

Research: To investigate the potential epigenetic regulation of gene expression in LOAD, we evaluated the human frontal cortex of 11 cases of LOAD and 12 cognitively normal controls. We measured genome-wide RNA gene expression using the Affymetrix U133A Plus 2.0 Array and DNA methylation using the Illumina HumanMethylation 27K BeadArray. We performed validation of X specific loci.

Results: After adjusting for age and sex, there were 176 probe sets (145 unique genes) distinguishing gene expression of LOAD cases and controls (FDR<0.1, p- value<3.3x10-4). Of these, 76.7% were down-regulated in LOAD, including the statistically top five genes DUSP16, ERICH1, ESF1, PTPRF, and RNBP1. Among the genes where expression was associated with LOAD (p-value<0.05), DNA methylation and gene expression were correlated at 151 genes. Positive (47.7%) and inverse (52.3%) associations were observed between gene expression and DNA methylation.

115

Conclusions: LOAD case and control samples differed by RNA expression levels and relatively few (7.2%) of those differences were correlated with DNA methylation. Among genes with correlated DNA methylation and gene expression, positive and negative associations were observed with approximately equal frequencies. Integrated genome-wide analyses of DNA methylation and RNA gene expression provide a functional molecular signature of LOAD and suggest novel sites for disease biomarker development.

Keywords: Alzheimer’s disease, epigenetics, DNA methylation, frontal cortex, gene expression

INTRODUCTION

Alzheimer’s disease (AD) is a fatal neurodegenerative disease that affects over five million people in the United States. It is the sixth leading cause of death across all ages and the prevalence is rising (Alzheimer's Association 2011). AD is characterized by two neuropathologies: β-amyloid (Aβ) plaques and tau neurofibrillary tangles (NFT). Highly penetrant genetic mutations in the Aβ production pathway account for approximately 2% of AD cases, termed early- onset AD (EOAD) cases (Bird 2005). The remaining 98% of AD cases manifest symptoms after age 55 and are termed late-onset AD (LOAD) cases. Genome- wide association studies reveal genetic risk factors for LOAD that are neither necessary nor sufficient to cause disease. LOAD risk is associated with polymorphisms in the apolipoprotein E4 (APOE ε4), ABCA7, BIN1, CD2AP, CD33, CLU, CR1, EPHA1, MS4A4A and PICALM genes (Harold et al. 2009; Hollingworth et al. 2011; Lambert et al. 2009; Naj et al. 2011). The combined population attributable fraction (PAF) for these multiple genetic risk factors is 0.50, even after adjusting for APOE ε4 dose (Naj et al. 2011). Thus, LOAD risk may be conferred through an interaction of external factors with genetic risk factors through gene-environment interactions or epigenetic modifications.

116

In LOAD, where genetic determinants confer only partial risk, Aβ and NFT pathways may be dysregulated through epigenetic mechanisms. Epigenetics refers to heritable changes in gene expression that do not involve changes to the underlying DNA sequence. Recent evidence suggests age-related changes in candidate gene methylation, such as upstream of Microtubule Associated Protein Tau (MAPT) in post-mortem human brain tissue (Tohgi et al. 1999). Striking disease-specific differences however, have not been consistently detected in promoter CpG island regions of Aβ and NFT candidate genes (Barrachina and Ferrer 2009). Newer research investigates non-canonical disease pathways, including cellular protein translation. For example, AD samples from the parietal and prefrontal cortex were hypermethylated in the promoter regions of nucleolar rRNA genes (Pietrzak et al. 2011), coinciding with reduced ribosomal activity that has been observed with LOAD.

To better understand the etiology of LOAD and the pathogenic processes, mapping the more complete epigenetic signature of disease is an important emerging area of research (Bakulski et al. 2012a). Recent work from our research group has broadened the targeted pathway approach to examine genome-wide DNA methylation in LOAD (Bakulski et al. 2012a; Bakulski et al. 2012b). DNA methylome studies in LOAD brain tissues suggest widespread, yet modest DNA methylation changes associated with both age and disease (Bakulski et al. 2012b). DNA methylation change is apparent, but the functional implications of the observed epigenetic marks have not yet been determined.

LOAD cases and controls have been classified based on their differential gene expression profiles in post-mortem brain tissues. For example, studies of gene expression by brain region in AD indicate significant dysregulation beyond the Aβ and NFT pathways (Kong et al. 2011). Expression levels of ceramide fatty acid processing genes including ASMas, NSMass2, and GALC were upregulated in the brains of AD subjects (Filippov et al. 2012). Changes in gene expression may also be at the center of hippocampal-dependent memory and corresponding deficiencies in AD. For example, hippocampal neurons derived

117

from AD transgenic mice had reduced expression of activity-induced CREB- dependent genes, which was reversible with expression of CREB-regulated transcription coactivator 1 (Crtc1) (Saura 2012). Transcriptome profiles of AD and controls suggest AD brain tissues have increased expression of broad markers of chronic inflammation, cell adhesion, cell proliferation, and protein synthesis, as well as downregulation of signal transduction, energy metabolism, stress response, synaptic vesicle synthesis and function, calcium binding, and structural proteins (Loring et al. 2001). The factors responsible for these broad gene expression changes are not clear. In order to understand the etiology of LOAD, we move upstream in the LOAD course and investigate the regulation of the LOAD transcriptome, controlled in part through the epigenome.

Recent individual gene studies have underscored the strengths in analyzing matched epigenetic and gene expression data in AD. The candidate gene, peptidyl-prolyl cis/trans isomerase (Pin1), was examined in human peripheral blood mononuclear cells of 60 matched LOAD and control samples (Arosio et al. 2012). Pin1 promoter DNA methylation had a modest reduction (8%) and increased gene expression (74%) in LOAD (Arosio et al. 2012). A second example involves the Aβ A4 precursor protein-binding family A member 2 (APBA2) gene implicated in Aβ production. In primary rat cortical neurons, expression of Apba2 was reduced with promoter DNA methylation at a location 7-120 base pairs upstream of the transcription start site (TSS) (Hao et al. 2012). An integrated analysis of observed DNA methylation and previously published RNA-seq expression was performed on a human cell line expressing the Swedish mutation of APP (APP-sw) and it revealed three potential gene sites (CTIF, NXT2, and DDR2) where DNA methylation and RNA expression were correlated (Sung et al. 2011). These results are promising, but cell lines have distinct epigenetic profiles from primary tissue (Smiraglia et al. 2001) and cell lines engineered with EOAD mutations are not adequate models for spontaneous LOAD. Previous studies show that analyses incorporating both epigenetic and gene expression data can be very useful (Sartor et al. 2011). Research using

118

primary human brain tissue in LOAD is needed to characterize the connection between DNA methylation and gene expression on a genome-wide scale.

Here we utilize whole genome approaches to interrogate both gene expression and DNA methylation profiles in LOAD cases and controls to identify potential functional relationships between epigenetic patterns and disease status. First, human post-mortem frontal cortex samples from LOAD and neuropathologically normal control subjects are assessed for gene expression profiles. Next, we combined gene expression data with DNA methylation results in the same samples. Identification of genes with disease-specific gene expression differences that are correlated with DNA methylation changes support an emerging role of epigenetics in LOAD.

Results

Gene expression results.

Gene expression in the frontal cortex was measured in 11 LOAD cases and 12 cognitively normal controls using the Affymetrix GeneChip Human Genome U133 Plus 2.0 Array. The array features 54,675 probe sets representing 20,722 unique gene ids. The mean expression level across LOAD and controls was 4.73 and it ranged from 1.75 to 14.25.

After adjusting for age and sex, site-specific expression values were compared between cases and controls. The DUSP16 (dual specificity phosphatase 16) gene transcript distinguished LOAD cases and controls with the greatest statistical significance (p-value= 8.85*10-7, FDR= 0.0316, log(2) fold change: -0.913). DUSP16 was down-regulated in LOAD, as were the next eight statistically significant genes with strongest association with LOAD status (Figure 4.1). These top nine genes (DUSP16, ERICH1, ESF1, PTPRF, FNBP1, RBM4, JMJD1C, LUC7L3, and RBM25) were down-regulated in AD case samples with an average of -1.18 log(2) fold change. Three of the top nine genes were RNA splicing regulators (RBM4, LUC7L3, RBM25).

119

The top 25 probesets that differentiated cases and controls at a false discovery rate of 0.033 are listed in Table 4.1. Twenty-two of the top 25 genes (88%) (DUSP16, ERICH1, ESF1, PTPRF, FNBP1, RBM4, JMJD1C, LUC7L3, RBM25, NUPL1, TCF25, UBXN4, CDC42BPA, CCAR1, ZNF37A, BRD4, EIF5B, LIMA1, ZFHX4, SMARCA4 , MYO6, CHAF1A) were down-regulated in LOAD. Three of the top 5 genes (12%) (SYTL4, DPYSL3, and LOC100505875) were up- regulated in LOAD.

Overall, there were 176 probe sets (146 unique genes) distinguishing LOAD cases and controls at a false discovery rate of less than 0.1 (p- value<3.2x10-4) (Supplemental Table 4.1). The log(2)fold change between cases and controls in these probe sets ranged from -1.98 to 1.72. Forty-one (23.9%) of these probe sets were up-regulated in LOAD, while 135 (76.7%) were down-regulated in LOAD. The normalized expression of the top 176 sites is illustrated in a heatmap of using Ward’s hierarchical clustering methods and the maximum distance function (Figure 4.2). Cases and controls are separated into two column clusters and the probeset rows separate into two groups based on up and down regulation with disease. These 176 probesets also distinguish cases and controls when using Principal Component Analysis. Principal component analysis (PCA) is a dimension reduction tool to reduce multiple correlated variables to linearly uncorrelated variables. When data were restricted to probe sets that were differentially expressed according to LOAD status (n=176), PCA was an effective visualization strategy. The first principal component described 67.2% of the variance and the second principal component described 9.5% of the variance (Figure 4.3A). PCA on the 176 probe sets associated with LOAD (FDR <0.1), clustered samples by LOAD case and control (Figure 4.3B). The heatmap and the PCA plot illustrate that these 176 probesets together are markers of LOAD case/control status in these samples.

To examine whether these differences observed between AD cases and controls were due to disease-specific changes in expression or changing cell- type populations, we incorporated publicly available cell-type specific data from

120

the NCBI Gene Expression Omnibus (GEO). We combined GEO data from embryonic stem cell-derived neuronal precursor cells (NPC) and astrocytes with our own AD case and control data. We used a follow-up heatmap to plot the expression levels at the original 176 probesets we found to distinguish cases and controls alongside the expression at those sites in NPC and astrocyte cells (Supplemental Figure 4.1). At these 176 sites, NPC and astrocyte cells were separated into two cell-type clusters. NPC and astrocytes did not cluster with AD cases or controls however, and the embryonic stem cell-derived data displayed greater range in gene expression change than our data.

To test whether there were biological patterns in the group of genes that were found to be different in expression between LOAD cases and controls, we used the software, LRpath, developed by the National Center for Integrative Biomedical Informatics (NCIBI) (Sartor et al. 2009). LRpath is a program that runs gene set enrichment testing using logistic regression on the full list of genes and associated p-values. We used LRpath to calculate overrepresentation among our top expression hits in gene ontology terms (biological processes and molecular functions), cytogenic bands, transcription factor targets, and miRNA targets. We observed down-regulation of 92 gene ontology biological processes at p<0.01 (FDR<0.1). The top fifteen biological processes that are decreased in LOAD are listed in Table 4.2 and the full list is in Supplemental Table 4.2. Nine processes were at false discovery rates < 0.002: synaptic transmission, transmission of nerve impulses, cell-cell signaling, oxidative phosphorylation, regulation of synaptic plasticity, regulation of transmission of nerve impulse, cellular respiration, regulation of neurological system process, neurotransmitter transport, and regulation of neuronal synaptic plasticity. Similarly, we observed enrichment of 348 gene ontology biological processes in the genes with increased expression in AD cases (p-value<0.01, FDR<0.1). The top fifteen biological processes that are increased in LOAD are listed in Table 4.3 and the full list is in Supplemental Table 4.3. The top nine were translational elongation, defense response, immune response, innate immune response, inflammatory response, response to biotic stimulus, response to other organism, activation of

121

immune response, and regulation of immune response at false discovery rates < 4*10-7. The results for gene ontology molecular functions were similar to the results for 77 significantly up and 79 down regulated biological processes (Supplemental Tables 4.4 and 4.5, respectively).

Genes differentially expressed in LOAD were also enriched in specific transcription factor targets, microRNA targets, and cytogenic bands. Reduced- expression in LOAD was associated with 17 transcription factors (Supplemental Table 4.6) including the top five: NRSF_01, NMYC_01, CREB_01, CREBP1CJUN_01, and MEIS1BHOXA9_02. Over-expression in LOAD was associated with the following six transcription factors NFKAPPAB_01, NKX22_01, OCT_C, PBX1_01, HOXA3_01, and CREL_01 (Supplemental Table 4.7). Genes targets associated with three microRNAs were over- expressed in LOAD (mir-506, mir-124, mir-433) and genes under-expressed in LOAD were enriched in binding for three microRNAs (mir-129-5p, mir-185, and mir-328) (Supplemental Tables 4.8 and 4.9, respectively). Among genes with higher expression in LOAD, 32 cytogenic bands were enriched, including the top five: 7p13-p12, 17p13, 4q25, 10p12.31, and 1q25 (Supplemental Table 4.10). Across genes with lower expression in LOAD, 33 cytogenic bands were enriched, including the top five: 16p13.3, 10p11.2, 8p23, 12p12.3, 8p21.3 (Supplemental Table 4.11).

Expression-Methylation

To test the extent to which DNA methylation was associated with the observed LOAD gene expression differences, we linked genome-wide gene expression and DNA methylation data from the same samples. DNA methylation data was generated using the Illumina Infinium HumanMethylation27 BeadArray (Bakulski et al. 2012b). We assessed combined methylation and expression at 2,094 Entrez gene IDs (Figure 4.4). These 2,094 genes had RNA expression associated with LOAD status (p-value <0.05) and DNA methylation data was available.

122

Across these 2,094 genes, we tested whether DNA methylation was correlated with gene expression. At 151 genes (7.2% of genes tested), DNA methylation and gene expression were significantly correlated (Pearson correlation test p-value<0.05). Among the 151 correlated sites, 79 (52.3%) represented the canonical, inverse association between DNA methylation and gene expression. Conversely, 72 (47.7%) represented non-canonical, positive correlation between DNA methylation and gene expression. Pearson correlation values ranged from -0.62 to 0.68 across these significantly correlated sites. Among the 151 DNA methylation and RNA expression correlated sites, 81 (53.6%) genes had higher methylation in LOAD and 70 (46.4%) had lower methylation in LOAD (Figure 4.5A). Similarly, 86 (57%) had higher expression in LOAD and 65 (43%) had lower expression (Figure 4.5B)

Significant differences in methylation levels between LOAD cases and controls were observed at 24 of the 151 DNA methylation-gene expression correlated sites (15.9%), after adjusting for age and sex (p-value<0.05) (Table 4.4). These 24 sites had significant disease-specific differences in DNA methylation, gene expression, and correlated methylation-expression values. Canonical, inverse correlation was observed at the following 17 genes (70.8% of 24 genes with significant methylation differences): ARHGAP15, PTAFR, FAM122C, SNX20, MKNK1, PSEN1, PMP2, CDC42EP3, PLD5, SKI, GPR34, TP53TG5, WWTR1, CATSPERG, PARVG, PPP1R3B, and IFI16. Non- canonical, positive correlation between DNA methylation and gene expression was observed at 7 (29.2%) of sites: TMPO, SERPINH1, PSMB2, ITGAM, MED12, OLFML2B, and NACC2. The first 9 genes with significant disease- specific differences in DNA methylation, gene expression, and correlated methylation-expression are illustrated in scatter plots in Figure 4.6 and the remaining 15 genes are in Supplemental Figure 4.2.

To test whether the genes with correlated expression and methylation values shared common biological pathways, we again performed gene set enrichment analysis using LRpath software. We uploaded the Pearson’s

123

correlation test p-values corresponding to the correlations between the 2,094 genes with linked RNA expression and DNA methylation data. We observed enrichment at the cytogenic bands across DNA methylation and RNA expression correlated genes: 17p11.2, 14q24.3, 16q22.1, 11q23.3, 6p21.2, and 19p13.3, as shown in the circos plot (Figure 4.7). We also observed transcription factor binding site enrichment at CEBPA_01, OCT1_05, GFI1_01, EVI1_06, MEF2_02, OCT1_01, NCX_01, PAX4_03, USF_C, GATA1_01, HAND1E47_01, PAX3_01, FOXO3_01, and COUP_01. The following microRNA targets were enriched among the correlated DNA methylation and RNA expression sites: mir-342-3p, mir-215, mir-192, mir-455-5p, mir-300, mir-488, and mir-142-3p. 171 biological processes were enriched, including the top ranked twelve: protein deubiquitination, protein modification by small protein removal, NK-kappB import into nucleus, regulation of NF-kappaB import into nucleus, negative regulation of axonogenesis, ion membrane transport, regulation of transcription factor import into nucleus, microtubule depolymerization, positive regulation of protein ubiquitination, nuclear-transcribed mRNA catabolic process, mRNA catabolic process, and positive regulation of ligase activity.

Discussion

LOAD is a complex disease involving multiple physiological changes that are evident in post-mortem pathology. Known genetic risk factors are responsible for only approximately half of the risk of LOAD disease initiation and progression (Naj et al. 2011). We interrogated the gene expression changes associated with disease and an epigenetic gene expression regulator. DNA methylation changes have been observed in LOAD, but the function of DNA methylation as a biomarker or potential mechanism of disease with gene expression implications had not been well characterized. We compared frontal cortex genome-wide RNA expression and DNA methylation profiles of LOAD cases and neuropathology- confirmed controls. Tissues were acquired on autopsy and nucleic acid samples were analyzed using array-based methods. We identified genes where expression was correlated with DNA methylation and where there were

124

differences by disease status. These observations support a potential functional role for DNA methylation in LOAD at a subset of genes.

First, using frontal cortex tissue, we identified discordant gene expression in LOAD. We identified a subset of genes that were associated with LOAD where expression of samples clustered by case status. These genes also differentiated NPC and astrocyte cell lines, which clustered separately from post- mortem samples. The majority of genes altered in LOAD were repressed in the disease samples. Across the entire gene set, the biological processes synaptic plasticity and transmission of nerve impulse were down-regulated in LOAD, while immune-related responses were up-regulated in LOAD. These findings are consistent with the current understanding of LOAD pathogenesis. In a novel analysis of genes with altered expression in LOAD, we integrated gene expression and DNA methylation data to test the role of DNA methylation in disease RNA changes. Seven percent of genes with altered expression in LOAD were correlated with DNA methylation levels. Gene expression varies with disease and DNA methylation may regulate the expression changes at a subset of genes.

A handful of identified genes have previously been implicated in LOAD and in cognitive changes. Specifically, we found reduced gene expression of ERICH1 in LOAD. Copy number variation within the ERICH1 gene was previously associated with intra-extradimensional set shifting (IED) domain on the CANTAB cognition test (Need et al. 2009). We also observed downregulation of formin binding protein 1 (FNBP1) in LOAD samples. FNBP1 was previously shown to be downregulated in blood mononuclear cells of AD subjects (Maes et al. 2007). FNBP1 is implicated in endocytosis and cellular trafficking and protein processing. Further, we observed reduced expression of PTPRF. AD pathogenesis is associated with abnormal autophagy and PTPRF binds PTPσ, an autophagic phosphatase (Martin et al. 2011).

125

The gene expression probe set which best distinguished LOAD cases and controls was associated with DUSP16 (dual specificity phosphatase 16). DUSP16 is a member of a family of proteins that catalyse the inactivation of mitogen-activated protein kinase (MAPK). MAPK plays a role in long term potentiation (LTP) of neurons (English and Sweatt 1997). The p38 MAPK pathway is involved in tau pathology in AD (Johnson and Bailey 2003) and has been targeted for AD treatment (Munoz and Ammit 2010). In addition, RBM4 was downregulated in LOAD in our study. RBM4 is a splice factor that interacts with an intronic element for MAPT and stimulates tau exon 10 inclusion (Kar et al. 2006). RMB25 and LUC7L3 are also splice factors that were downregulated in LOAD in our study. Dysregulation of splicing factors could have widespread downstream effects for neurodegenerative disease (Licatalosi and Darnell 2006).

Previous research has performed transcriptome array analysis on LOAD cases and controls in the neocortex and found a large number of genes with divergent disease gene expression with wide-ranging physiological functions (Tan et al. 2010). Among the genes that were associated with LOAD (FDR<0.1), we observed 77% were down-regulated in gene expression with LOAD. Similarly, Maes et al. found of the 942 genes with 2-fold differential change in AD blood mononuclear cells relative to controls, 87% were downregulated and only 13% were upregulated in AD (Maes et al. 2007). Research has looked at genetics as a driver of gene expression differences. A previous genome-wide gene expression study in Alzheimer’s disease linked SNP data with transcript levels (Webster et al. 2009), and found that ~5% of transcripts had correlation between expression profiles and genotypes that could distinguish LOAD cases and controls. Results from our study suggest that a similar level of genes with a altered expression (7% of genes) are correlated with DNA methylation levels. The current paper is the first study, that we are aware of, to examine the association between DNA methylation and gene expression in LOAD on a genome-wide scale. Future research may integrate genetics, transcriptomics, and epigenomics for a more complete understanding of gene regulation in LOAD.

126

A limitation of cross-sectional research, such as this, is that causal relationships cannot be inferred. Later work will be able to study the time course of the DNA methylation and gene expression associations observed here. Further, LOAD is characterized by neuronal loss (Coleman and Flood 1987) and glial cell activation (Mattson 2004). Changing cell type populations may be responsible for these observations and future work could identify the cell type specific DNA methylation and gene expression patterns.

GWAS studies estimate that known genetic LOAD risk alleles confer a combined population attributable fraction of 0.5 (Naj et al. 2011). Additional risk genotypes may be discovered, but the remaining risk likely results from environmental exposures and an interaction of genetic and environmental factors, which have been historically understudied. Epigenetics represents an important intersection of genes and environment. By identifying new candidate areas of epigenetic and gene expression variability with LOAD, this research provides the groundwork for future studies in environmental and genetic susceptibility to LOAD. LOAD research may increasingly consider a combined gene-environment paradigm.

Materials and Methods

Sample ascertainment. The Michigan Alzheimer’s Disease Center (MADC) (P50AG008671; PI: Henry Paulson) followed a clinically characterized cohort of Alzheimer’s disease and cognitively normal control subjects. Many subjects (and legal care-givers) consented to autopsy and donated to the MADC Brain Bank. Diagnoses were neuropathologically confirmed using Braak and Reagan scoring in the left hemispheres. The right hemispheres were coronally sectioned, flash frozen, and archived at -80ºC. For the current study, frozen tissue blocks from the mid-frontal gyrus of the frontal lobe were dissected at -20ºC. MADC frozen tissues were previously used in high quality DNA methylation (Bakulski et al. 2012b), expression (Hong et al. 2008; Pan et al. 2007), and proteomic studies (Pan et al. 2007).

127

Samples were eligible for the current study if neuropathology confirmed the diagnosis and if the subjects were at least 60 years old at death. The MADC Brain Bank had 98 eligible LOAD cases (Braak Score ≥ 4) and 39 controls (Braak Score ≤ 2). Twelve gender- and age- (+/- 2 years) matched pairs of LOAD cases and controls were generated randomly, without replication. The population characteristics of both sets of samples used in the current study are described in Table 4.5. This human subjects study was approved by the Institutional Review Board of the University of Michigan Medical School.

DNA and RNA isolation. Gray matter free of vascular lesions was selected from the tissue blocks. DNA and RNA were isolated from adjoining tissues in all 25 matched pairs. DNA was extracted according to manufacturer’s instructions using the Maxwell Tissue DNA Kit (Promega, AS1030). Tissue was homogenized with the TissueLyser II (Qiagen, 85300) and total RNA was extracted using the RNeasy Lipid Tissue Mini Kit (Qiagen, 74804).

Transcriptome-wide RNA expression. RNA from the twelve case-control pairs was interrogated for genome-wide expression analysis. RNA quality was assessed using the Agilent 2100 Bioanalyzer and quantified using the Thermo Scientific NanoDrop Spectrophotometer 2000. Samples with RNA quality greater than 3.0 RNA Integrity Number (RIN) and absorbance ratios A260/A280 greater than 2.0 were used for further study. One AD case sample did not meet this set of criteria and was not used for further RNA study. The quality of RNA did not differ by LOAD cases and controls (mean RIN=6.0, p-value=0.8). 5 ug of RNA was used for cDNA synthesis using a T7-Oligo(dT) promoter primer. An in vitro transcription (IVT) reaction produced biotin-labeled cRNA that was then fragmented. At the University of Michigan Affymetrix Core facility, cRNA was hybridized to the GeneChip Human Genome U133 Plus 2.0 Array (Affymetrix, 900470) using the 16-hour hybridization protocol. Each sample was hybridized to its own chip. The probe array was washed and stained according to the GeneChip Expression Wash, Stain and Scan User Manual (Affymetrix, 702731). The probe array was scanned using the Affymetrix Scanner 3000 instrument and

128

the image files were analyzed for probe intensities. The GeneChip provided data on 20,722 unique genes using 54,675 probe sets.

Affymetrix Data Processing. All further data processing and statistical analyses were done in R Statistical Software (version 2.18). Data was background corrected, log2 transformed, and quantile normalized with Robust Multichip Average (RMA) methods (Gautier et al. 2004). Using the affyQCReport and simpleaffy packages, quality control between the chips was high.

Gene specific differential expression and significance by AD case status was assessed using parallel linear models and empirical Bayesian variance methods at site-specific moderated t-tests (Smyth 2004). The following parallel linear models were fit at each CpG site that adjusted for subject’s age and sex.

Expression = β0 + β1(AD Case Status) +β2(Age) + β3(Sex)

Probesets with AD Case Status β values that are significant at false discovery <0.1 are listed in a table and are used for data visualization. A heatmap with mean normalized expression values was used to cluster samples and probesets. To further illustrate differences between gene expression profiles in LOAD cases and controls, principle component analysis was used. Across all probes on the GeneChip, the first three principal components explain 67% of variance.

For each probeset, the Entrez gene ID, corresponding AD Case Status β p-value, and T-test statistic for direction of association were uploaded to the LRpath gene set enrichment testing program (Sartor et al. 2009). LRpath was used to test for enrichment in gene ontology terms, cytogenic bands, transcription factor binding sites, and microRNA binding sites.

We also downloaded publicly available cell-type specific gene expression data from NCBI’s Gene Expression Omnibus (GEO). Gene expression in embryonic stem cell derived neuronal precursor cells (GSE7178) and astrocytes

129

(GSE5080) were plotted in a heatmap alongside the top hits distinguishing LOAD cases and controls.

Genome-wide DNA methylation. One µg of DNA from the twelve case-control pairs in the Discovery Phase was bisulfite-treated according to manufacturer’s instructions with the EZ DNA Methylation Kit (Zymo, D5001). The Illumina recommended extended bisulfite thermocycling protocol (98ºC 10 minutes, hold at 64ºC for 17 hours) was followed. Bisulfite-treated DNA was prepped and applied to the Infinium HumanMethylation27 BeadArray (Illumina, WG-311-2201) by the University of Michigan DNA Sequencing Core facility using methods published by Illumina researchers (Bibikova et al. 2009). Across 27,578 probe sets, the BeadChip primarily targets CpG sites in the promoter regions of 14,475 genes and 110 miRNAs. Batch effects can potentially bias experimental differences, so six cases and six control samples were randomly applied to each of the 12-sample BeadChips. Fluorescent intensities were imaged using the Illumina BeadArray Reader and the associated BeadScan software was used for image processing and data extraction. Data were background normalized and percent methylation estimates (beta values) were calculated for each probe set. Data were exported to R for further processing.

Illumina Data Processing. The Illumina GenomeStudio software scored CpG sites for individual samples as failing based on fluorescence. CpG sites that failed on greater than 10% of samples (n=171 sites) were excluded from analyses. Tests for differences in methylation by epidemiological characteristics were performed using the limma BioConductor R package. The following parallel linear models were fit at each CpG site that adjusted for subject’s age and sex.

% Methylation = β0 + β1(AD Case Status) +β2(Age) + β3(Sex)

Data integration and statistical methods. We filtered the list of genes differentially expressed between LOAD cases and controls and selected only the probe sets that met an expression difference threshold of p<0.05 (n=4,063 probe sets). We removed 750 probe sets that did not map to Entrez gene ID’s

130

(n=3,313). To reduce the data to a single expression value per unique gene ID, if greater than one probe set was differentially expressed for a given gene, we selected the probe with the highest expression value. The set of unique Affymetrix gene ID’s that were associated with LOAD case status included 2,768 probe sets.

We filtered the Illumina methylation data to only include CpG sites within 1500 bp of a transcription start site with a known Entrez gene ID (n=25,811). In the case of multiple Illumina probes for a given gene ID, we selected the CpG site with the lowest p-value for the association between LOAD cases and controls (n=13,865).

We then matched the Affymetrix expression data and the Illumina methylation data by Entrez gene IDs. This yielded a total of 2,094 unique gene ID’s where Illumina and Affymetrix matched and where the Affymetrix probe set was statistically significantly associated with LOAD case status (Figure 4.4). The Pearson’s correlation between gene expression and DNA methylation was calculated for each gene. We performed LRpath gene set enrichment analysis on the 2,094 gene ID’s with linked expression and methylation data.

131

FIGURES

Figure 4.1. Boxplots of gene expression of the 9 top genes that differ by LOAD case or control status. LOAD cases display reduced gene expression at all of 9 of the genes.

132

Figure 4.2. Heatmap of all of the 176 sites on the Affymetrix gene expression array (FDR < 0.1) associated with LOAD case status. This plot used maximum distance and Ward’s hierarchical clustering methods. Data has been normalized to the mean expression value per probeset.

133

A. B.

Figure 4.3. Principal Component Analysis. (A) Principal component loading histogram. (B). Principal component analysis scree plot for the top 176 probesets (FDR < 0.1).

134

Figure 4.4. Gene-expression and DNA methylation linked data analysis pipeline.

A. Figure 4.5 (A) Scatterplot of the 133 genes that displayed discordant gene expression and DNA methylation between LOAD cases and controls. The top of the figure represents genes less expressed in AD (n=50) and the bottom of the figure represents genes upregulated (n=84) in LOAD. The left of the figure shows genes more highly methylated in LOAD and the right shows less

135

methylation in LOAD. (B) Boxplot of the differences in methylation between AD and controls separated by the direction of change in expression.

A. B. C.

D. E. F.

G. H. I.

Figure 4.6. Scatter plots of 9 genes: expression vs. methylation.

136

Figure 4.7. Circos plot: Locations of gene expression change and DNA methylation change.

137

TABLES

Table 4.1. Affymetrix expression differences between LOAD cases and controls (n=25) after adjusting for age and sex. Rank Affymetrix ID Entrez ID Gene Symbol Gene Name Chromosome Average Fold Log2 Fold T-test P-value Adjusted Location Expression Change Change Statistic P-value 1 1558740_s_at 80824 DUSP16 dual specificity phosphatase 16 12p13.2 7.11 0.531 -0.913 -6.63 8.85E-07 0.0316 2 227016_at 157697 ERICH1 glutamate-rich 1 8p23.3 5.37 0.529 -0.917 -6.26 2.09E-06 0.0316 3 218859_s_at 51575 ESF1 ESF1, nucleolar pre-rRNA processing protein, 20p12.1 7.72 0.393 -1.349 -6.16 2.65E-06 0.0316 homolog (S. cerevisiae) 4 200635_s_at 5792 PTPRF protein tyrosine phosphatase, receptor type, F 1p34 7.35 0.488 -1.034 -6.12 2.96E-06 0.0316 5 213940_s_at 23048 FNBP1 formin binding protein 1 9q34 7.03 0.425 -1.233 -5.87 5.33E-06 0.0316 6 213718_at 5936 RBM4 RNA binding motif protein 4 11q13 5.86 0.537 -0.897 -5.85 5.57E-06 0.0316 7 224933_s_at 221037 JMJD1C jumonji domain containing 1C 10q21.3 7.19 0.504 -0.990 -5.83 5.96E-06 0.0316 8 208835_s_at 51747 LUC7L3 LUC7-like 3 (S. cerevisiae) 17q21.33 10.14 0.399 -1.327 -5.78 6.59E-06 0.0316 9 212027_at 58517 RBM25 RNA binding motif protein 25 14q24.3 7.73 0.264 -1.919 -5.74 7.27E-06 0.0316 10 241425_at 9818 NUPL1 nucleoporin like 1 13q12.13 5.70 0.558 -0.842 -5.73 7.60E-06 0.0316 11 213311_s_at 22980 TCF25 transcription factor 25 (basic helix-loop-helix) 16q24.3 8.47 0.374 -1.420 -5.71 7.89E-06 0.0316

138 12 212007_at 23190 UBXN4 UBX domain protein 4 2q21.3 8.97 0.421 -1.247 -5.66 8.88E-06 0.0316 13 214464_at 8476 CDC42BPA CDC42 binding protein kinase alpha (DMPK-like) 1q42.11 8.32 0.303 -1.722 -5.64 9.27E-06 0.0316

Table 4.2. Among genes with lower gene expression in AD cases vs. controls, the following are the top 15 biological processes that are down-regulated based on LR-Path. Rank Name #Genes P-Value FDR SigGenes 1 synaptic transmission 353 4.92E-12 2.92E-09 223, 5368, 627, 6529, 7425, 9379 2 transmission of nerve impulse 411 6.52E-12 3.38E-09 223, 5368, 627, 6529, 7425, 9379 3 cell-cell signaling 700 5.95E-07 6.03E-05 223, 4826, 5368, 627, 6529, 7425, 9379, 9547, 9636 4 oxidative phosphorylation 94 3.10E-06 1.81E-04 155066, 4704 5 regulation of synaptic plasticity 53 8.57E-06 4.23E-04 627, 7425 6 regulation of transmission of nerve impulse 132 2.83E-05 1.10E-03 627, 6529, 7425 7 cellular respiration 94 2.90E-05 1.10E-03 4704 8 regulation of neurological system process 142 4.41E-05 1.57E-03 627, 6529, 7425 9 neurotransmitter transport 96 5.54E-05 1.93E-03 6529, 9379 10 regulation of neuronal synaptic plasticity 32 5.96E-05 2.06E-03 7425 11 regulation of synaptic transmission 121 7.69E-05 0.003 627, 6529, 7425 12 respiratory electron transport chain 59 8.08E-05 0.003 4704 13 regulation of neurotransmitter levels 82 1.13E-04 0.003 223, 9379 14 nucleotide-excision repair, DNA damage removal 21 1.16E-04 0.003 2073 15 ATP synthesis coupled electron transport 51 1.89E-04 0.005 4704 139

Table 4.3. Among genes with higher gene expression in AD cases vs. controls, the following are the top 15 biological processes that are up-regulated based on LR-Path. Rank Name #Genes P-Value FDR SigGenes 1 translational elongation 101 4.07E-20 1.69E-16 11224, 1937, 2197, 25873, 6135, 6154, 6173, 6176, 6188, 6193, 6207, 6222, 6232 10219, 10410, 1050, 10581, 11326, 12, 23643, 241, 2532, 2919, 313, 3440, 3487, 3588, 3600, 3823, 4057, 4615, 4688, 51191, 58191, 6039, 604, 60675, 6283, 7097, 2 defense response 661 1.16E-16 2.42E-13 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 7305, 8519, 929, 9332, 9450 10346, 10410, 10581, 11326, 23643, 2669, 2919, 3108, 3109, 3122, 3588, 3600, 4057, 4615, 4688, 4860, 51191, 54209, 563, 58191, 604, 6398, 7097, 7098, 710, 3 immune response 660 3.86E-15 5.34E-12 7100, 712, 713, 714, 718, 719, 8519, 929, 9450 10219, 1050, 11326, 12, 1462, 2162, 23643, 241, 2697, 2919, 313, 3399, 3440, 3487, 3587, 3588, 3600, 4615, 4814, 604, 60675, 6283, 7097, 7098, 710, 7100, 712, 4 response to wounding 598 1.07E-14 1.11E-11 713, 7132, 714, 718, 719, 7423, 929, 9332, 9450 10410, 10581, 11326, 23643, 4615, 4688, 51191, 58191, 7097, 7098, 710, 7100, 5 innate immune response 201 8.28E-13 6.68E-10 712, 713, 714, 718, 8519, 929, 9450 10219, 1050, 11326, 12, 23643, 241, 2919, 313, 3440, 3487, 3588, 3600, 4615, 604, 6 inflammatory response 365 9.65E-13 6.68E-10 60675, 6283, 7097, 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 929, 9332, 9450 10049, 10346, 10410, 10581, 11080, 1373, 1937, 23643, 3315, 3440, 3587, 3600, 7 response to biotic stimulus 442 8.34E-12 3.85E-09 3665, 3669, 4057, 51191, 6283, 7079, 7097, 7098, 7100, 7132, 8519, 871, 929

140 10346, 10410, 10581, 1373, 1937, 23643, 3315, 3440, 3587, 3600, 3665, 3669, 8 response to other organism 354 5.69E-11 2.36E-08 4057, 51191, 6283, 7079, 7097, 7098, 7100, 7132, 8519, 929

9 activation of immune response 110 7.45E-11 2.81E-08 11326, 1997, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719 10 regulation of immune response 245 8.89E-10 3.08E-07 11326, 1997, 3600, 604, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719 11 response to bacterium 213 1.42E-09 4.53E-07 1373, 23643, 3587, 4057, 6283, 7079, 7097, 7098, 7100, 7132, 929 12 I-kappaB kinase/NF-kappaB cascade 171 2.35E-09 6.98E-07 23643, 2697, 3965, 4615, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132 regulation of toll-like receptor signaling 13 pathway 9 2.97E-09 8.21E-07 7097, 7098, 7100 positive regulation of toll-like receptor 14 signaling pathway 7 7.11E-09 1.85E-06 7097, 7098, 7100 15 humoral immune response 80 8.56E-09 2.09E-06 11326, 4057, 54209, 710, 712, 713, 714, 718, 9450

Table 4.4. Gene expression and methylation correlation.

Rank Entrez ID Gene Symbol Affymetrix ID Illumina ID Log2 Expression Percent Methylation Pearson Pearson Expression Difference P- Difference in Difference P- Correlation Correlation Fold Change Value Methylation Value Coefficient Test P-Value 1 7112 TMPO 209754_s_at cg23264519 0.309 0.041 0.411 0.046 0.617 0.002 2 55843 ARHGAP15 218870_at cg23627134 0.457 0.003 -3.241 0.016 -0.614 0.002 3 5724 PTAFR 227184_at cg18468844 0.436 0.011 -3.088 0.031 -0.591 0.003 4 159091 FAM122C 239047_at cg16510010 -0.259 0.003 4.715 0.016 -0.586 0.003 5 124460 SNX20 228869_at cg27081230 0.261 0.016 -4.881 0.016 -0.567 0.005 6 8569 MKNK1 1560720_at cg15445332 0.230 0.002 -0.362 0.029 -0.540 0.008 7 5663 PSEN1 207782_s_at cg11490446 0.430 0.008 -1.044 0.034 -0.534 0.009 8 871 SERPINH1 207714_s_at 6.250186 0.878 0.041 1.507 0.026 0.525 0.010 9 5375 PMP2 206826_at cg21649520 -1.124 0.000 4.849 0.024 -0.521 0.011 10 5690 PSMB2 231323_at 3.890297 0.402 0.027 0.606 0.007 0.509 0.013 11 3684 ITGAM 205786_s_at cg00833777 0.509 0.030 7.910 0.004 0.503 0.015 12 10602 CDC42EP3 209288_s_at cg05469695 0.446 0.040 -1.993 0.027 -0.491 0.017 13 200150 PLD5 1563933_a_at cg12613383 0.657 0.002 -2.358 0.042 -0.488 0.018

141 14 6497 SKI 213755_s_at cg06382459 -0.263 0.032 1.217 0.011 -0.484 0.019 15 2857 GPR34 223620_at cg22835805 0.670 0.031 -5.528 0.006 -0.464 0.026

16 27296 TP53TG5 207482_at cg14226064 -0.407 0.029 3.634 0.006 -0.461 0.027 17 25937 WWTR1 202133_at cg12507125 0.725 0.010 -2.123 0.004 -0.454 0.030 18 9968 MED12 203506_s_at cg21693321 0.242 0.037 5.736 0.029 0.449 0.032 19 57828 CATSPERG 231261_at cg18996334 -0.189 0.042 0.234 0.043 -0.448 0.032 20 64098 PARVG 233510_s_at cg19863740 0.385 0.004 -2.099 0.035 -0.445 0.033 21 25903 OLFML2B 213125_at cg20172280 0.242 0.043 0.803 0.022 0.443 0.034 22 79660 PPP1R3B 222662_at cg24727203 0.601 0.004 -7.281 0.011 -0.441 0.035 23 138151 NACC2 212993_at cg12004206 0.421 0.022 0.358 0.044 0.441 0.035 24 3428 IFI16 208966_x_at cg21406461 0.598 0.030 -7.925 0.039 -0.438 0.036

Table 4.5. Study population characteristics.

LOAD Cases Neuropathologically Normal Controls N 11 12 Age mean(range) 80 (69-94) 79.9 (69-95) Sex 6 Females 6 Females 5 Males 6 Males Ethnicity Caucasian Caucasian Braak Stage mean 4.75 1.33 Years in storage mean (range) 11 (5-17) 13.2 (3-20)

142

REFERENCES

Alzheimer's Association I. 2011. Alzheimer's disease facts and figures. Alzheimer's & Dementia 7(2). Arosio B, Bulbarelli A, Bastias Candia S, Lonati E, Mastronardi L, Romualdi P, Candeletti S, Gussago C, Galimberti D, Scarpini E, Dell'osso B, Altamura C, Maccarrone M, Bergamaschini L, D'Addario C, Mari D. 2012. Pin1 Contribution to Alzheimer's Disease: Transcriptional and Epigenetic Mechanisms in Patients with Late-Onset Alzheimer's Disease. Neurodegener Dis. Bakulski KM, Dolinoy DC, Paulson HL, Rozek LS, Hu H. 2012a. Alzheimer's disease and environmental exposure to lead: The epidemiologic evidence and potential role of epigenetics. Current Alzheimer's Research (In Press). Bakulski KM, Dolinoy DC, Sartor MA, Paulson HL, Lieberman AP, Konen JR, Albin RL, Hu H, Rozek LS. 2012b. Genome-wide DNA methlyation differences between late-onset Alzheimer's disease and cognitively normal controls in human frontal cortex. Journal of Alzheimer's Disease 29(2). Barrachina M, Ferrer I. 2009. DNA methylation of Alzheimer disease and tauopathy-related genes in postmortem brain. J Neuropathol Exp Neurol 68(8): 880-891. Bibikova M, Le J, Barnes B, Saedinia-Melnyk S, Zhou LX, Shen R, Gunderson KL. 2009. Genome-wide DNA methylation profiling using Infinium (R) assay. Epigenomics-Uk 1(1): 177-200. Bird TD. 2005. Genetic factors in Alzheimer's disease. N Engl J Med 352(9): 862- 864. Coleman PD, Flood DG. 1987. Neuron numbers and dendritic extent in normal aging and Alzheimer's disease. Neurobiol Aging 8(6): 521-545. English JD, Sweatt JD. 1997. A requirement for the mitogen-activated protein kinase cascade in hippocampal long term potentiation. J Biol Chem 272(31): 19103-19106. Filippov V, Song MA, Zhang K, Vinters HV, Tung S, Kirsch WM, Yang J, Duerksen-Hughes PJ. 2012. Increased Ceramide in Brains with Alzheimer's and other Neurodegenerative Diseases. J Alzheimers Dis. Gautier L, Cope L, Bolstad BM, Irizarry RA. 2004. affy--analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20(3): 307-315. Hao Y, Chai KH, McLoughlin DM, Chan HY, Lau KF. 2012. Promoter characterization and genomic organization of the human X11beta gene APBA2. Neuroreport.

143

Harold D, Abraham R, Hollingworth P, Sims R, Gerrish A, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Williams A, Jones N, Thomas C, Stretton A, Morgan AR, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC, Gill M, Lawlor B, Lynch A, Morgan K, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Love S, Kehoe PG, Hardy J, Mead S, Fox N, Rossor M, Collinge J, Maier W, Jessen F, Schurmann B, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Hull M, Rujescu D, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Carrasquillo MM, Pankratz VS, Younkin SG, Holmans PA, O'Donovan M, Owen MJ, Williams J. 2009. Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer's disease. Nat Genet 41(10): 1088-1093. Hollingworth P, Harold D, Sims R, Gerrish A, Lambert JC, Carrasquillo MM, Abraham R, Hamshere ML, Pahwa JS, Moskvina V, Dowzell K, Jones N, Stretton A, Thomas C, Richards A, Ivanov D, Widdowson C, Chapman J, Lovestone S, Powell J, Proitsi P, Lupton MK, Brayne C, Rubinsztein DC, Gill M, Lawlor B, Lynch A, Brown KS, Passmore PA, Craig D, McGuinness B, Todd S, Holmes C, Mann D, Smith AD, Beaumont H, Warden D, Wilcock G, Love S, Kehoe PG, Hooper NM, Vardy ER, Hardy J, Mead S, Fox NC, Rossor M, Collinge J, Maier W, Jessen F, Ruther E, Schurmann B, Heun R, Kolsch H, van den Bussche H, Heuser I, Kornhuber J, Wiltfang J, Dichgans M, Frolich L, Hampel H, Gallacher J, Hull M, Rujescu D, Giegling I, Goate AM, Kauwe JS, Cruchaga C, Nowotny P, Morris JC, Mayo K, Sleegers K, Bettens K, Engelborghs S, De Deyn PP, Van Broeckhoven C, Livingston G, Bass NJ, Gurling H, McQuillin A, Gwilliam R, Deloukas P, Al-Chalabi A, Shaw CE, Tsolaki M, Singleton AB, Guerreiro R, Muhleisen TW, Nothen MM, Moebus S, Jockel KH, Klopp N, Wichmann HE, Pankratz VS, Sando SB, Aasly JO, Barcikowska M, Wszolek ZK, Dickson DW, Graff-Radford NR, Petersen RC, van Duijn CM, Breteler MM, Ikram MA, Destefano AL, Fitzpatrick AL, Lopez O, Launer LJ, Seshadri S, Berr C, Campion D, Epelbaum J, Dartigues JF, Tzourio C, Alperovitch A, Lathrop M, Feulner TM, Friedrich P, Riehle C, Krawczak M, Schreiber S, Mayhaus M, Nicolhaus S, Wagenpfeil S, Steinberg S, Stefansson H, Stefansson K, Snaedal J, Bjornsson S, Jonsson PV, Chouraki V, Genier-Boley B, Hiltunen M, Soininen H, Combarros O, Zelenika D, Delepine M, Bullido MJ, Pasquier F, Mateo I, Frank-Garcia A, Porcellini E, Hanon O, Coto E, Alvarez V, Bosco P, Siciliano G, Mancuso M, Panza F, Solfrizzi V, Nacmias B, Sorbi S, Bossu P, Piccardi P, Arosio B, Annoni G, Seripa D, Pilotto A, Scarpini E, Galimberti D, Brice A, Hannequin D, Licastro F, Jones L, Holmans PA, Jonsson T, Riemenschneider M, Morgan K, Younkin SG, Owen MJ, O'Donovan M,

144

Amouyel P, Williams J. 2011. Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer's disease. Nat Genet. Hong MG, Myers AJ, Magnusson PK, Prince JA. 2008. Transcriptome-wide assessment of human brain and lymphocyte senescence. PLoS One 3(8): e3024. Johnson GV, Bailey CD. 2003. The p38 MAP kinase signaling pathway in Alzheimer's disease. Exp Neurol 183(2): 263-268. Kar A, Havlioglu N, Tarn WY, Wu JY. 2006. RBM4 interacts with an intronic element and stimulates tau exon 10 inclusion. J Biol Chem 281(34): 24479-24488. Kong W, Mou X, Hu X. 2011. Exploring matrix factorization techniques for significant genes identification of Alzheimer's disease microarray gene expression data. BMC Bioinformatics 12 Suppl 5: S7. Lambert JC, Heath S, Even G, Campion D, Sleegers K, Hiltunen M, Combarros O, Zelenika D, Bullido MJ, Tavernier B, Letenneur L, Bettens K, Berr C, Pasquier F, Fievet N, Barberger-Gateau P, Engelborghs S, De Deyn P, Mateo I, Franck A, Helisalmi S, Porcellini E, Hanon O, de Pancorbo MM, Lendon C, Dufouil C, Jaillard C, Leveillard T, Alvarez V, Bosco P, Mancuso M, Panza F, Nacmias B, Bossu P, Piccardi P, Annoni G, Seripa D, Galimberti D, Hannequin D, Licastro F, Soininen H, Ritchie K, Blanche H, Dartigues JF, Tzourio C, Gut I, Van Broeckhoven C, Alperovitch A, Lathrop M, Amouyel P. 2009. Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer's disease. Nat Genet 41(10): 1094-1099. Licatalosi DD, Darnell RB. 2006. Splicing regulation in neurologic disease. Neuron 52(1): 93-101. Loring JF, Wen X, Lee JM, Seilhamer J, Somogyi R. 2001. A gene expression profile of Alzheimer's disease. DNA Cell Biol 20(11): 683-695. Maes OC, Xu S, Yu B, Chertkow HM, Wang E, Schipper HM. 2007. Transcriptional profiling of Alzheimer blood mononuclear cells by microarray. Neurobiol Aging 28(12): 1795-1809. Martin KR, Xu Y, Looyenga BD, Davis RJ, Wu CL, Tremblay ML, Xu HE, MacKeigan JP. 2011. Identification of PTPsigma as an autophagic phosphatase. J Cell Sci 124(Pt 5): 812-819. Mattson MP. 2004. Pathways towards and away from Alzheimer's disease. Nature 430(7000): 631-639.

145

Munoz L, Ammit AJ. 2010. Targeting p38 MAPK pathway for the treatment of Alzheimer's disease. Neuropharmacology 58(3): 561-568. Naj AC, Jun G, Beecham GW, Wang LS, Vardarajan BN, Buros J, Gallins PJ, Buxbaum JD, Jarvik GP, Crane PK, Larson EB, Bird TD, Boeve BF, Graff- Radford NR, De Jager PL, Evans D, Schneider JA, Carrasquillo MM, Ertekin-Taner N, Younkin SG, Cruchaga C, Kauwe JS, Nowotny P, Kramer P, Hardy J, Huentelman MJ, Myers AJ, Barmada MM, Demirci FY, Baldwin CT, Green RC, Rogaeva E, George-Hyslop PS, Arnold SE, Barber R, Beach T, Bigio EH, Bowen JD, Boxer A, Burke JR, Cairns NJ, Carlson CS, Carney RM, Carroll SL, Chui HC, Clark DG, Corneveaux J, Cotman CW, Cummings JL, Decarli C, Dekosky ST, Diaz-Arrastia R, Dick M, Dickson DW, Ellis WG, Faber KM, Fallon KB, Farlow MR, Ferris S, Frosch MP, Galasko DR, Ganguli M, Gearing M, Geschwind DH, Ghetti B, Gilbert JR, Gilman S, Giordani B, Glass JD, Growdon JH, Hamilton RL, Harrell LE, Head E, Honig LS, Hulette CM, Hyman BT, Jicha GA, Jin LW, Johnson N, Karlawish J, Karydas A, Kaye JA, Kim R, Koo EH, Kowall NW, Lah JJ, Levey AI, Lieberman AP, Lopez OL, Mack WJ, Marson DC, Martiniuk F, Mash DC, Masliah E, McCormick WC, McCurry SM, McDavid AN, McKee AC, Mesulam M, Miller BL, Miller CA, Miller JW, Parisi JE, Perl DP, Peskind E, Petersen RC, Poon WW, Quinn JF, Rajbhandary RA, Raskind M, Reisberg B, Ringman JM, Roberson ED, Rosenberg RN, Sano M, Schneider LS, Seeley W, Shelanski ML, Slifer MA, Smith CD, Sonnen JA, Spina S, Stern RA, Tanzi RE, Trojanowski JQ, Troncoso JC, Van Deerlin VM, Vinters HV, Vonsattel JP, Weintraub S, Welsh-Bohmer KA, Williamson J, Woltjer RL, Cantwell LB, Dombroski BA, Beekly D, Lunetta KL, Martin ER, Kamboh MI, Saykin AJ, Reiman EM, Bennett DA, Morris JC, Montine TJ, Goate AM, Blacker D, Tsuang DW, Hakonarson H, Kukull WA, Foroud TM, Haines JL, Mayeux R, Pericak-Vance MA, Farrer LA, Schellenberg GD. 2011. Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer's disease. Nat Genet. Need AC, Attix DK, McEvoy JM, Cirulli ET, Linney KL, Hunt P, Ge D, Heinzen EL, Maia JM, Shianna KV, Weale ME, Cherkas LF, Clement G, Spector TD, Gibson G, Goldstein DB. 2009. A genome-wide study of common SNPs and CNVs in cognitive performance in the CANTAB. Hum Mol Genet 18(23): 4650-4661. Pan S, Shi M, Jin J, Albin RL, Lieberman A, Gearing M, Lin B, Pan C, Yan X, Kashima DT, Zhang J. 2007. Proteomics identification of proteins in human cortex using multidimensional separations and MALDI tandem mass spectrometer. Mol Cell Proteomics 6(10): 1818-1823. Pietrzak M, Rempala G, Nelson PT, Zheng JJ, Hetman M. 2011. Epigenetic silencing of nucleolar rRNA genes in Alzheimer's disease. PLoS One 6(7): e22585.

146

Sartor MA, Dolinoy DC, Jones TR, Colacino JA, Prince ME, Carey TE, Rozek LS. 2011. Genome-wide methylation and expression differences in HPV(+) and HPV(-) squamous cell carcinoma cell lines are consistent with divergent mechanisms of carcinogenesis. Epigenetics 6(6): 777-787. Sartor MA, Leikauf GD, Medvedovic M. 2009. LRpath: a logistic regression approach for identifying enriched biological groups in gene expression data. Bioinformatics 25(2): 211-217. Saura CA. 2012. CREB-Regulated Transcription Coactivator 1-Dependent Transcription in Alzheimer's Disease Mice. Neurodegener Dis. Smiraglia DJ, Rush LJ, Fruhwald MC, Dai Z, Held WA, Costello JF, Lang JC, Eng C, Li B, Wright FA, Caligiuri MA, Plass C. 2001. Excessive CpG island hypermethylation in cancer cell lines versus primary human malignancies. Hum Mol Genet 10(13): 1413-1419. Smyth GK. 2004. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3: Article3. Sung HY, Choi EN, Ahn Jo S, Oh S, Ahn JH. 2011. Amyloid protein-mediated differential DNA methylation status regulates gene expression in Alzheimer's disease model cell line. Biochem Biophys Res Commun 414(4): 700-705. Tan MG, Chua WT, Esiri MM, Smith AD, Vinters HV, Lai MK. 2010. Genome wide profiling of altered gene expression in the neocortex of Alzheimer's disease. J Neurosci Res 88(6): 1157-1169. Tohgi H, Utsugisawa K, Nagane Y, Yoshimura M, Genda Y, Ukitsu M. 1999. Reduction with age in methylcytosine in the promoter region -224 approximately -101 of the amyloid precursor protein gene in autopsy human cortex. Brain Res Mol Brain Res 70(2): 288-292. Webster JA, Gibbs JR, Clarke J, Ray M, Zhang W, Holmans P, Rohrer K, Zhao A, Marlowe L, Kaleem M, McCorquodale DS, 3rd, Cuello C, Leung D, Bryden L, Nath P, Zismann VL, Joshipura K, Huentelman MJ, Hu-Lince D, Coon KD, Craig DW, Pearson JV, Heward CB, Reiman EM, Stephan D, Hardy J, Myers AJ. 2009. Genetic control of human brain transcript expression in Alzheimer disease. Am J Hum Genet 84(4): 445-458.

147

SUPPLEMENTAL FIGURES

Supplemental Figure 4.1. Heatmap of all of the 176 sites on the Affymetrix gene expression array (FDR < 0.1) associated with LOAD case status. Publicly available data from embryonic stem cell derived neuronal precursor cells (GSE7178) and astrocytes (GSE5080) have been included. This plot used maximum distance and Ward’s hierarchical clustering methods and normalized to the mean expression value per probeset.

148

A. B. C.

D. E. F.

G. H. I.

J. K. L.

M. N. O.

149

P. Q. R.

S. T. U.

V. W. X.

Supplemental Figure 2. Scatterplots of expression vs. methylation for the 23 genes significantly associated with AD via expression (p<0.05) and methylation (p<0.05), and methylation and expression are significantly correlated (Pearson’s p<0.05).

150

SUPPLEMENTAL TABLES

Supplemental Table 4.1. Gene expression probesets associated with AD case vs. controls at FDR <0.1 (n=176) Rank Affymetrix ID Entrez ID Gene Symbol Gene Name Chromosome Average Fold Log2 Fold T-test P-value Adjusted Location Expression Change Change Statistic P-value 1 1558740_s_at 80824 DUSP16 dual specificity phosphatase 16 12p13.2 7.11 0.531 -0.913 -6.63 8.85E-07 0.0316 2 227016_at 157697 ERICH1 glutamate-rich 1 8p23.3 5.37 0.529 -0.917 -6.26 2.09E-06 0.0316 3 218859_s_at 51575 ESF1 ESF1, nucleolar pre-rRNA processing protein, 20p12.1 7.72 0.393 -1.349 -6.16 2.65E-06 0.0316 homolog (S. cerevisiae) 4 200635_s_at 5792 PTPRF protein tyrosine phosphatase, receptor type, F 1p34 7.35 0.488 -1.034 -6.12 2.96E-06 0.0316 5 213940_s_at 23048 FNBP1 formin binding protein 1 9q34 7.03 0.425 -1.233 -5.87 5.33E-06 0.0316 6 213718_at 5936 RBM4 RNA binding motif protein 4 11q13 5.86 0.537 -0.897 -5.85 5.57E-06 0.0316 7 224933_s_at 221037 JMJD1C jumonji domain containing 1C 10q21.3 7.19 0.504 -0.990 -5.83 5.96E-06 0.0316 8 208835_s_at 51747 LUC7L3 LUC7-like 3 (S. cerevisiae) 17q21.33 10.14 0.399 -1.327 -5.78 6.59E-06 0.0316 9 212027_at 58517 RBM25 RNA binding motif protein 25 14q24.3 7.73 0.264 -1.919 -5.74 7.27E-06 0.0316 10 241425_at 9818 NUPL1 nucleoporin like 1 13q12.13 5.70 0.558 -0.842 -5.73 7.60E-06 0.0316 11 213311_s_at 22980 TCF25 transcription factor 25 (basic helix-loop-helix) 16q24.3 8.47 0.374 -1.420 -5.71 7.89E-06 0.0316

151 12 212007_at 23190 UBXN4 UBX domain protein 4 2q21.3 8.97 0.421 -1.247 -5.66 8.88E-06 0.0316 13 214464_at 8476 CDC42BPA CDC42 binding protein kinase alpha (DMPK-like) 1q42.11 8.32 0.303 -1.722 -5.64 9.27E-06 0.0316

14 227703_s_at 94121 SYTL4 synaptotagmin-like 4 Xq21.33 6.31 2.538 1.344 5.64 9.32E-06 0.0316 15 224736_at 55749 CCAR1 cell division cycle and apoptosis regulator 1 10q21.3 8.02 0.446 -1.165 -5.64 9.47E-06 0.0316 16 228711_at 7587 ZNF37A zinc finger protein 37A 10p11.2 7.10 0.584 -0.775 -5.61 1.01E-05 0.0316 17 201430_s_at 1809 DPYSL3 dihydropyrimidinase-like 3 5q32 5.01 1.937 0.954 5.61 1.01E-05 0.0316 18 226054_at 23476 BRD4 bromodomain containing 4 19p13.1 7.45 0.538 -0.894 -5.59 1.05E-05 0.0316 19 201026_at 9669 EIF5B eukaryotic translation initiation factor 5B 2q11.2 7.24 0.465 -1.104 -5.58 1.10E-05 0.0316 20 230781_at 100505875 LOC100505875 uncharacterized LOC100505875 NA 4.56 2.494 1.318 5.52 1.26E-05 0.0322 21 222457_s_at 51474 LIMA1 LIM domain and actin binding 1 12q13 5.52 0.432 -1.212 -5.51 1.28E-05 0.0322 22 219779_at 79776 ZFHX4 zinc finger homeobox 4 8q21.11 6.64 0.627 -0.674 -5.50 1.33E-05 0.0322 23 212520_s_at 6597 SMARCA4 SWI/SNF related, matrix associated, actin 19p13.2 7.19 0.427 -1.229 -5.49 1.35E-05 0.0322 dependent regulator of chromatin, subfamily a, member 4 24 203215_s_at 4646 MYO6 myosin VI 6q13 7.26 0.253 -1.984 -5.47 1.41E-05 0.0322 25 203975_s_at 10036 CHAF1A chromatin assembly factor 1, subunit A (p150) 19p13.3 4.89 0.674 -0.569 -5.45 1.51E-05 0.0330 26 232595_at NA NA NA NA 3.49 1.360 0.444 5.41 1.66E-05 0.0349 27 222628_s_at 51455 REV1 REV1 homolog (S. cerevisiae) 2q11.1-q11.2 7.19 0.530 -0.916 -5.37 1.81E-05 0.0366 28 200842_s_at 2058 EPRS glutamyl-prolyl-tRNA synthetase 1q41 7.79 0.367 -1.444 -5.36 1.88E-05 0.0366 29 222792_s_at 29080 CCDC59 coiled-coil domain containing 59 12q21.31 6.50 0.566 -0.822 -5.34 1.98E-05 0.0366 30 214375_at 8496 PPFIBP1 PTPRF interacting protein, binding protein 1 (liprin 12p12.1 6.27 0.440 -1.185 -5.33 2.03E-05 0.0366 beta 1) 31 220727_at 54207 KCNK10 potassium channel, subfamily K, member 10 14q31.3 6.30 0.577 -0.794 -5.30 2.16E-05 0.0366 32 206726_at 27306 HPGDS hematopoietic prostaglandin D synthase 4q22.3 3.32 1.353 0.436 5.29 2.23E-05 0.0366 33 242233_at NA NA NA NA 5.52 0.603 -0.730 -5.28 2.25E-05 0.0366 34 202379_s_at 4820 NKTR natural killer-tumor recognition sequence 3p23-p21 9.16 0.476 -1.070 -5.28 2.28E-05 0.0366

35 214911_s_at 6046 BRD2 bromodomain containing 2 6p21.3 8.26 0.467 -1.098 -5.25 2.47E-05 0.0386 36 216563_at 23253 ANKRD12 ankyrin repeat domain 12 18p11.22 7.50 0.389 -1.363 -5.22 2.64E-05 0.0399 37 211993_at 65125 WNK1 WNK lysine deficient protein kinase 1 12p13.3 6.85 0.344 -1.540 -5.20 2.76E-05 0.0399 38 219387_at 55704 CCDC88A coiled-coil domain containing 88A 2p16.1 8.08 0.286 -1.807 -5.16 3.05E-05 0.0399 39 226176_s_at 84132 USP42 ubiquitin specific peptidase 42 7p22.1 6.36 0.510 -0.971 -5.16 3.09E-05 0.0399 40 223185_s_at 79365 BHLHE41 basic helix-loop-helix family, member e41 12p12.1 4.14 0.464 -1.108 -5.15 3.10E-05 0.0399 41 209945_s_at 2932 GSK3B glycogen synthase kinase 3 beta 3q13.3 5.25 0.703 -0.509 -5.14 3.19E-05 0.0399 42 229353_s_at 64710 NUCKS1 nuclear casein kinase and cyclin-dependent kinase 1q32.1 10.51 0.526 -0.926 -5.14 3.20E-05 0.0399 substrate 1 43 225041_at 54737 MPHOSPH8 M-phase phosphoprotein 8 13q12.11 8.89 0.376 -1.410 -5.14 3.24E-05 0.0399 44 225565_at 1385 CREB1 cAMP responsive element binding protein 1 2q34 5.33 0.552 -0.857 -5.13 3.27E-05 0.0399 45 222122_s_at 57187 THOC2 THO complex 2 Xq25-q26.3 7.11 0.462 -1.114 -5.13 3.28E-05 0.0399 46 213298_at 4782 NFIC nuclear factor I/C (CCAAT-binding transcription 19p13.3 5.44 0.447 -1.161 -5.10 3.55E-05 0.0422 factor) 47 209230_s_at 26471 NUPR1 nuclear protein, transcriptional regulator, 1 16p11.2 7.06 2.413 1.271 5.08 3.74E-05 0.0432 48 222540_s_at 51773 RSF1 remodeling and spacing factor 1 11q14.1 8.25 0.398 -1.328 -5.07 3.79E-05 0.0432 49 222620_s_at 64215 DNAJC1 DnaJ (Hsp40) homolog, subfamily C, member 1 10p12.31 7.29 0.385 -1.378 -5.04 4.08E-05 0.0442 50 1569594_a_at 9147 NEMF nuclear export mediator factor 14q22 7.97 0.429 -1.221 -5.04 4.08E-05 0.0442 51 229586_at 80205 CHD9 chromodomain helicase DNA binding protein 9 16q12.2 7.65 0.476 -1.072 -5.04 4.13E-05 0.0442

152 52 208963_x_at 3992 FADS1 fatty acid desaturase 1 11q12.2- 7.71 0.556 -0.847 -5.00 4.52E-05 0.0467 q13.1

53 226975_at 55599 RNPC3 RNA-binding region (RNP1, RRM) containing 3 1p21 6.65 0.393 -1.346 -5.00 4.53E-05 0.0467 54 229163_at 55450 CAMK2N1 calcium/calmodulin-dependent protein kinase II 1p36.12 7.38 0.567 -0.820 -4.99 4.69E-05 0.0472 inhibitor 1 55 1555495_a_at 10283 CWC27 CWC27 spliceosome-associated protein homolog 5q12.3 8.27 0.566 -0.820 -4.97 4.85E-05 0.0472 (S. cerevisiae) 56 200702_s_at 57062 DDX24 DEAD (Asp-Glu-Ala-Asp) box polypeptide 24 14q32 7.59 0.358 -1.480 -4.97 4.88E-05 0.0472 57 225946_at 11228 RASSF8 Ras association (RalGDS/AF-6) domain family (N- 12p12.3 6.58 1.665 0.736 4.96 5.04E-05 0.0472 terminal) member 8 58 209376_x_at 9169 SCAF11 SR-related CTD-associated factor 11 12q12 7.02 0.468 -1.095 -4.96 5.09E-05 0.0472 59 208994_s_at 9360 PPIG peptidylprolyl isomerase G (cyclophilin G) 2q31.1 7.81 0.406 -1.299 -4.95 5.10E-05 0.0472 60 213850_s_at 9169 SCAF11 SR-related CTD-associated factor 11 12q12 7.86 0.460 -1.122 -4.94 5.27E-05 0.0480 61 201183_s_at 1108 CHD4 chromodomain helicase DNA binding protein 4 12p13 7.21 0.461 -1.118 -4.93 5.35E-05 0.0480 62 232323_s_at 55761 TTC17 tetratricopeptide repeat domain 17 11p11.2 6.38 0.469 -1.092 -4.92 5.60E-05 0.0494 63 209088_s_at 29855 UBN1 ubinuclein 1 16p13.3 6.99 0.418 -1.257 -4.91 5.75E-05 0.0497 64 215049_x_at 9332 CD163 CD163 molecule 12p13.3 5.22 3.284 1.715 4.90 5.85E-05 0.0497 65 222616_s_at 10600 USP16 ubiquitin specific peptidase 16 21q22.11 6.39 0.329 -1.606 -4.89 5.92E-05 0.0497 66 213729_at 55660 PRPF40A PRP40 pre-mRNA processing factor 40 homolog A 2q23.3 6.87 0.460 -1.120 -4.89 5.99E-05 0.0497 (S. cerevisiae) 67 231061_at NA NA NA NA 5.42 0.540 -0.888 -4.88 6.11E-05 0.0499 68 217728_at 6277 S100A6 S100 calcium binding protein A6 1q21 8.34 1.537 0.620 4.87 6.26E-05 0.0500 69 203729_at 2014 EMP3 epithelial membrane protein 3 19q13.3 5.56 1.977 0.983 4.86 6.47E-05 0.0500 70 227298_at 401264 FLJ37798 uncharacterized LOC401264 6p12.3 6.06 0.750 -0.414 -4.86 6.48E-05 0.0500 71 229635_at 1.01E+08 LOC100505702 uncharacterized LOC100505702 NA 4.25 2.236 1.161 4.86 6.50E-05 0.0500

72 208685_x_at 6046 BRD2 bromodomain containing 2 6p21.3 8.27 0.487 -1.038 -4.84 6.73E-05 0.0511 73 217832_at 10492 SYNCRIP synaptotagmin binding, cytoplasmic RNA interacting 6q14-q15 7.35 1.764 0.819 4.82 7.16E-05 0.0536 protein 74 208772_at NA NA NA NA 8.03 0.579 -0.788 -4.80 7.56E-05 0.0557 75 219437_s_at 29123 ANKRD11 ankyrin repeat domain 11 16q24.3 6.84 0.376 -1.411 -4.79 7.75E-05 0.0557 76 238595_at NA NA NA NA 5.86 0.487 -1.037 -4.79 7.77E-05 0.0557 77 238584_at 79781 IQCA1 IQ motif containing with AAA domain 1 2q37.3 4.26 0.467 -1.099 -4.78 7.85E-05 0.0557 78 35436_at 2801 GOLGA2 golgin A2 9q34.11 6.16 0.495 -1.015 -4.76 8.20E-05 0.0569 79 1558965_at 51317 PHF21A PHD finger protein 21A 11p11.2 3.12 1.419 0.505 4.76 8.24E-05 0.0569 80 241458_at NA NA NA NA 4.63 1.432 0.518 4.76 8.38E-05 0.0569 81 220946_s_at 29072 SETD2 SET domain containing 2 3p21.31 3.30 0.550 -0.862 -4.75 8.53E-05 0.0569 82 214843_s_at 23032 USP33 ubiquitin specific peptidase 33 1p31.1 6.00 0.600 -0.738 -4.75 8.59E-05 0.0569 83 236869_at NA NA NA NA 4.33 0.553 -0.854 -4.74 8.74E-05 0.0569 84 207542_s_at 358 AQP1 aquaporin 1 (Colton blood group) 7p14 6.41 2.433 1.283 4.74 8.75E-05 0.0569 85 201914_s_at 11231 SEC63 SEC63 homolog (S. cerevisiae) 6q21 6.13 0.384 -1.380 -4.73 8.93E-05 0.0575 86 221210_s_at 80896 NPL N-acetylneuraminate pyruvate lyase 1q25 3.93 2.039 1.028 4.72 9.21E-05 0.0581 (dihydrodipicolinate synthase) 87 203181_x_at 6733 SRPK2 SRSF protein kinase 2 7q22-q31.1 9.55 0.496 -1.011 -4.71 9.43E-05 0.0581 88 233080_s_at 55660 PRPF40A PRP40 pre-mRNA processing factor 40 homolog A 2q23.3 6.77 0.490 -1.029 -4.70 9.70E-05 0.0581

153 (S. cerevisiae) 89 209466_x_at 5764 PTN pleiotrophin 7q33 8.46 0.411 -1.282 -4.69 9.81E-05 0.0581

90 208663_s_at 7267 TTC3 tetratricopeptide repeat domain 3 21q22.2 8.39 0.298 -1.745 -4.69 9.98E-05 0.0581 91 211737_x_at 5764 PTN pleiotrophin 7q33 9.37 0.445 -1.168 -4.68 1.02E-04 0.0581 92 214314_s_at 9669 EIF5B eukaryotic translation initiation factor 5B 2q11.2 6.17 0.551 -0.861 -4.68 1.02E-04 0.0581 93 206929_s_at 4782 NFIC nuclear factor I/C (CCAAT-binding transcription 19p13.3 7.19 0.499 -1.002 -4.67 1.03E-04 0.0581 factor) 94 209127_s_at 9733 SART3 squamous cell carcinoma antigen recognized by T 12q24.1 6.13 0.417 -1.260 -4.67 1.03E-04 0.0581 cells 3 95 212382_at 6925 TCF4 transcription factor 4 18q21.1 7.25 0.387 -1.371 -4.67 1.04E-04 0.0581 96 225590_at 57630 SH3RF1 SH3 domain containing ring finger 1 4q32.3 5.92 0.595 -0.749 -4.67 1.04E-04 0.0581 97 226782_at 253512 SLC25A30 solute carrier family 25, member 30 13q14.13 5.11 1.364 0.448 4.67 1.05E-04 0.0581 98 204964_s_at 8082 SSPN sarcospan (Kras oncogene-associated gene) 12p11.2 6.29 0.497 -1.008 -4.66 1.06E-04 0.0581 99 208942_s_at 7095 SEC62 SEC62 homolog (S. cerevisiae) 3q26.2 9.81 0.596 -0.746 -4.65 1.08E-04 0.0581 100 232677_at NA NA NA NA 4.55 0.603 -0.731 -4.65 1.09E-04 0.0581 101 208993_s_at 9360 PPIG peptidylprolyl isomerase G (cyclophilin G) 2q31.1 8.28 0.463 -1.111 -4.64 1.10E-04 0.0581 102 208610_s_at 23524 SRRM2 serine/arginine repetitive matrix 2 16p13.3 7.99 0.326 -1.619 -4.64 1.11E-04 0.0581 103 216520_s_at 7178 TPT1 tumor protein, translationally-controlled 1 13q14 11.68 1.383 0.468 4.64 1.12E-04 0.0581 104 201224_s_at 10250 SRRM1 serine/arginine repetitive matrix 1 1p36.11 8.23 0.543 -0.882 -4.64 1.12E-04 0.0581 105 225730_s_at 25917 THUMPD3 THUMP domain containing 3 3p25.3 5.40 0.463 -1.111 -4.64 1.12E-04 0.0581 106 242916_at 11064 CNTRL centriolin 9q33.2 4.22 0.575 -0.799 -4.64 1.13E-04 0.0581 107 212570_at 23052 ENDOD1 endonuclease domain containing 1 11q21 5.80 0.496 -1.012 -4.63 1.14E-04 0.0581 108 215338_s_at 4820 NKTR natural killer-tumor recognition sequence 3p23-p21 7.00 0.510 -0.970 -4.61 1.20E-04 0.0609 109 211996_s_at NA NA NA NA 10.44 0.621 -0.686 -4.60 1.24E-04 0.0615 110 230180_at 10521 DDX17 DEAD (Asp-Glu-Ala-Asp) box helicase 17 22q13.1 5.80 0.510 -0.971 -4.60 1.24E-04 0.0615

111 1569302_at 85459 KIAA1731 KIAA1731 11q21 5.21 0.508 -0.976 -4.59 1.27E-04 0.0625 112 208676_s_at 5036 PA2G4 proliferation-associated 2G4, 38kDa 12q13.2 7.93 0.718 -0.478 -4.59 1.28E-04 0.0625 113 218659_at 55252 ASXL2 additional sex combs like 2 (Drosophila) 2p24.1 6.85 1.483 0.568 4.57 1.32E-04 0.0634 114 211948_x_at 23215 PRRC2C proline-rich coiled-coil 2C 1q23.3 8.90 0.557 -0.843 -4.57 1.33E-04 0.0634 115 225377_at 55684 C9orf86 chromosome 9 open reading frame 86 9q34.3 7.78 0.408 -1.293 -4.57 1.33E-04 0.0634 116 1555913_at 54856 GON4L gon-4-like (C. elegans) 1q22 4.58 0.643 -0.636 -4.57 1.34E-04 0.0634 117 221745_at 10238 DCAF7 DDB1 and CUL4 associated factor 7 17q23.3 5.94 0.641 -0.642 -4.56 1.36E-04 0.0635 118 1556211_a_at NA NA NA NA 4.88 0.300 -1.737 -4.55 1.41E-04 0.0654 119 214305_s_at 23451 SF3B1 splicing factor 3b, subunit 1, 155kDa 2q33.1 6.57 0.510 -0.971 -4.54 1.44E-04 0.0663 120 209579_s_at 8930 MBD4 methyl-CpG binding domain protein 4 3q21.3 8.48 0.529 -0.918 -4.52 1.50E-04 0.0683 121 228801_at 94101 ORMDL1 ORM1-like 1 (S. cerevisiae) 2q32 4.16 0.678 -0.561 -4.51 1.53E-04 0.0692 122 239894_at 1E+08 LOC100128511 uncharacterized LOC100128511 10p12.31 3.94 0.614 -0.703 -4.51 1.55E-04 0.0694 123 202845_s_at 10928 RALBP1 ralA binding protein 1 18p11.3 8.73 1.381 0.466 4.50 1.58E-04 0.0696 124 201085_s_at 6651 SON SON DNA binding protein 21q22.1- 7.75 0.356 -1.492 -4.50 1.58E-04 0.0696 q22.2 125 206826_at 5375 PMP2 peripheral myelin protein 2 8q21.3-q22.1 10.04 0.459 -1.124 -4.47 1.70E-04 0.0739 126 201024_x_at 9669 EIF5B eukaryotic translation initiation factor 5B 2q11.2 9.48 0.524 -0.934 -4.47 1.70E-04 0.0739 127 203186_s_at 6275 S100A4 S100 calcium binding protein A4 1q21 5.35 1.685 0.753 4.46 1.74E-04 0.0742 128 242835_s_at 728730 LOC728730 uncharacterized LOC728730 2p22.1 5.29 0.380 -1.395 -4.46 1.75E-04 0.0742 154 129 239154_at NA NA NA NA 5.39 0.578 -0.790 -4.46 1.76E-04 0.0742 130 232617_at 1520 CTSS cathepsin S 1q21 5.85 1.904 0.929 4.46 1.77E-04 0.0742

131 203761_at 6503 SLA Src-like-adaptor 8q22.3-qter 6.22 1.909 0.933 4.45 1.78E-04 0.0742 132 238893_at 338758 LOC338758 uncharacterized LOC338758 12q21.33 7.45 0.506 -0.983 -4.45 1.79E-04 0.0742 133 204999_s_at 22809 ATF5 activating transcription factor 5 19q13.3 3.64 0.670 -0.577 -4.44 1.84E-04 0.0758 134 241955_at 25831 HECTD1 HECT domain containing E3 ubiquitin protein ligase 14q12 5.29 0.413 -1.275 -4.43 1.89E-04 0.0770 1 135 213509_x_at 8824 CES2 carboxylesterase 2 16q22.1 7.20 0.716 -0.482 -4.42 1.92E-04 0.0778 136 218454_at 79887 PLBD1 phospholipase B domain containing 1 12p13.1 4.02 1.365 0.449 4.42 1.95E-04 0.0786 137 209258_s_at 9126 SMC3 structural maintenance of chromosomes 3 10q25 5.46 0.386 -1.372 -4.41 2.00E-04 0.0794 138 201730_s_at 7175 TPR translocated promoter region (to activated MET 1q25 6.69 0.426 -1.233 -4.41 2.01E-04 0.0794 oncogene) 139 231729_s_at 828 CAPS calcyphosine 19p13.3 4.69 2.011 1.008 4.40 2.02E-04 0.0794 140 1552326_a_at 220136 CCDC11 coiled-coil domain containing 11 18q21.1 4.19 1.442 0.528 4.39 2.11E-04 0.0822 141 208095_s_at NA NA NA NA 6.95 0.353 -1.502 -4.38 2.12E-04 0.0822 142 208879_x_at 24148 PRPF6 PRP6 pre-mRNA processing factor 6 homolog (S. 20q13.33 6.57 0.430 -1.218 -4.38 2.16E-04 0.0832 cerevisiae) 143 212994_at 57187 THOC2 THO complex 2 Xq25-q26.3 6.17 0.506 -0.984 -4.37 2.22E-04 0.0850 144 244154_at 80821 DDHD1 DDHD domain containing 1 14q21 4.89 0.722 -0.471 -4.36 2.25E-04 0.0853 145 222020_s_at 50863 NTM neurotrimin 11q25 6.37 0.508 -0.977 -4.35 2.33E-04 0.0863 146 204787_at 11326 VSIG4 V-set and immunoglobulin domain containing 4 Xq12-q13.3 5.92 2.774 1.472 4.34 2.35E-04 0.0863 147 209715_at 23468 CBX5 chromobox homolog 5 12q13.13 7.32 0.564 -0.825 -4.34 2.36E-04 0.0863 148 242974_at 961 CD47 CD47 molecule 3q13.1-q13.2 5.50 1.343 0.426 4.34 2.38E-04 0.0863 149 214129_at 9659 PDE4DIP phosphodiesterase 4D interacting protein 1q12 6.28 1.945 0.960 4.34 2.38E-04 0.0863 150 208930_s_at 3609 ILF3 interleukin enhancer binding factor 3, 90kDa 19p13.2 5.53 0.452 -1.147 -4.34 2.39E-04 0.0863

151 219507_at 51319 RSRC1 arginine/serine-rich coiled-coil 1 3q25.32 5.74 0.360 -1.473 -4.33 2.41E-04 0.0863 152 223138_s_at 170506 DHX36 DEAH (Asp-Glu-Ala-His) box polypeptide 36 3p13-q23 5.53 0.471 -1.085 -4.33 2.41E-04 0.0863 153 222737_s_at 29117 BRD7 bromodomain containing 7 16q12 7.03 0.513 -0.962 -4.33 2.42E-04 0.0863 154 1554470_s_at 29068 ZBTB44 zinc finger and BTB domain containing 44 11q24.3 4.40 1.524 0.608 4.33 2.43E-04 0.0864 155 226416_at 90459 ERI1 exoribonuclease 1 8p23.1 5.18 1.401 0.486 4.33 2.46E-04 0.0868 156 206167_s_at 395 ARHGAP6 Rho GTPase activating protein 6 Xp22.3 5.74 1.485 0.570 4.32 2.48E-04 0.0870 157 204061_at 5613 PRKX protein kinase, X-linked Xp22.3 5.49 2.377 1.249 4.32 2.51E-04 0.0870 158 202844_s_at 10928 RALBP1 ralA binding protein 1 18p11.3 7.35 0.470 -1.090 -4.32 2.51E-04 0.0870 159 203645_s_at 9332 CD163 CD163 molecule 12p13.3 4.50 2.827 1.499 4.30 2.59E-04 0.0892 160 224631_at NA NA NA NA 6.65 0.447 -1.162 -4.30 2.64E-04 0.0901 161 217869_at 51144 HSD17B12 hydroxysteroid (17-beta) dehydrogenase 12 11p11.2 9.37 0.726 -0.461 -4.29 2.70E-04 0.0918 162 235409_at 23269 MGA MAX gene associated 15q14 5.91 0.491 -1.026 -4.28 2.78E-04 0.0937 163 224856_at 2289 FKBP5 FK506 binding protein 5 6p21.31 6.81 1.951 0.964 4.27 2.81E-04 0.0940 164 212120_at 23433 RHOQ ras homolog family member Q 2p21 8.27 1.367 0.451 4.27 2.86E-04 0.0940 165 241769_at NA NA NA NA 5.38 0.541 -0.888 -4.27 2.86E-04 0.0940 166 226189_at 3696 ITGB8 integrin, beta 8 7p21.1 9.05 1.469 0.555 4.26 2.87E-04 0.0940 167 223797_at 114224 PRO2852 uncharacterized protein PRO2852 NA 6.19 1.467 0.553 4.26 2.87E-04 0.0940 168 208710_s_at 8943 AP3D1 adaptor-related protein complex 3, delta 1 subunit 19p13.3 8.12 0.627 -0.674 -4.26 2.91E-04 0.0946 169 214055_x_at 23215 PRRC2C proline-rich coiled-coil 2C 1q23.3 8.85 0.553 -0.855 -4.25 2.99E-04 0.0961 155 170 242728_at NA NA NA NA 4.54 1.474 0.560 4.25 2.99E-04 0.0961 171 224605_at 401152 C4orf3 chromosome 4 open reading frame 3 4q26 8.04 0.611 -0.711 -4.25 3.00E-04 0.0961 172 201606_s_at 11137 PWP1 PWP1 homolog (S. cerevisiae) 12q23.3 6.49 0.486 -1.040 -4.24 3.02E-04 0.0961 173 221043_at NA NA NA NA 3.57 0.621 -0.687 -4.23 3.12E-04 0.0987 174 213328_at 4750 NEK1 NIMA (never in mitosis gene a)-related kinase 1 4q33 6.24 0.463 -1.111 -4.23 3.16E-04 0.0987 175 227221_at 64393 ZMAT3 zinc finger, matrin-type 3 3q26.32 6.74 1.315 0.395 4.22 3.17E-04 0.0987 176 239946_at NA NA NA NA 4.77 1.382 0.467 4.22 3.18E-04 0.0987

Supplemental Table 4.2. Among genes with lower expression in Alzheimer’s, several biological processes were enriched.

Rank Name #Genes P-Value FDR SigGenes 1 synaptic transmission 353 4.92E-12 2.92E-09 223, 5368, 627, 6529, 7425, 9379 2 transmission of nerve impulse 411 6.52E-12 3.38E-09 223, 5368, 627, 6529, 7425, 9379 3 cell-cell signaling 700 5.95E-07 6.03E-05 223, 4826, 5368, 627, 6529, 7425, 9379, 9547, 9636 4 oxidative phosphorylation 94 3.10E-06 1.81E-04 155066, 4704 5 regulation of synaptic plasticity 53 8.57E-06 4.23E-04 627, 7425 6 regulation of transmission of nerve impulse 132 2.83E-05 1.10E-03 627, 6529, 7425 7 cellular respiration 94 2.90E-05 1.10E-03 4704 8 regulation of neurological system process 142 4.41E-05 1.57E-03 627, 6529, 7425 9 neurotransmitter transport 96 5.54E-05 1.93E-03 6529, 9379 10 regulation of neuronal synaptic plasticity 32 5.96E-05 2.06E-03 7425 11 regulation of synaptic transmission 121 7.69E-05 0.003 627, 6529, 7425 12 respiratory electron transport chain 59 8.08E-05 0.003 4704 13 regulation of neurotransmitter levels 82 1.13E-04 0.003 223, 9379 14 nucleotide-excision repair, DNA damage removal 21 1.16E-04 0.003 2073 15 ATP synthesis coupled electron transport 51 1.89E-04 0.005 4704 16 mitochondrial ATP synthesis coupled electron transport 51 1.89E-04 0.005 4704 17 generation of a signal involved in cell-cell signaling 178 1.92E-04 0.005 4826, 627, 7425, 9379

156 18 signal release 178 1.92E-04 0.005 4826, 627, 7425, 9379 19 oxygen transport 5 2.37E-04 0.006 3049 20 mitochondrial electron transport, NADH to ubiquinone 37 4.13E-04 0.01 4704

21 tRNA metabolic process 113 5.68E-04 0.013 26995, 54938, 80222 22 cerebellar cortex formation 9 8.19E-04 0.017 23 synaptic vesicle transport 31 1.12E-03 0.023 24 cellular amino acid metabolic process 227 1.17E-03 0.023 10157, 2954, 54938, 80222 25 ncRNA metabolic process 230 1.32E-03 0.026 26995, 5393, 54938, 6839, 80222 26 cerebellar cortex morphogenesis 12 1.50E-03 0.028 27 synaptic vesicle exocytosis 14 1.52E-03 0.029 28 nucleobase, nucleoside and nucleotide metabolic process 551 1.55E-03 0.029 155066, 254272, 3704, 482, 56342, 79077, 8382 29 aspartate family amino acid catabolic process 6 1.70E-03 0.03 10157 30 negative regulation of synaptic transmission, GABAergic 7 1.91E-03 0.033 6529 31 generation of precursor metabolites and energy 324 2.01E-03 0.033 155066, 4704, 7425 32 cellular amino acid catabolic process 68 2.32E-03 0.037 10157, 2954 33 neurotransmitter secretion 47 2.36E-03 0.038 9379 34 DNA dealkylation 6 2.40E-03 0.038 35 DNA dealkylation involved in DNA repair 6 2.40E-03 0.038 36 regulation of megakaryocyte differentiation 14 0.003 0.039 8364, 8366 37 fear response 18 0.003 0.039 627 38 L-amino acid import 8 0.003 0.041 6529 39 lysosome organization 24 0.003 0.043 53, 84067 40 nucleoside phosphate metabolic process 529 0.004 0.049 155066, 254272, 3704, 482, 56342, 79077, 8382 41 nucleotide metabolic process 529 0.004 0.049 155066, 254272, 3704, 482, 56342, 79077, 8382 42 ATP hydrolysis coupled proton transport 9 0.004 0.051 155066 43 energy coupled proton transport, against electrochemical gradient 9 0.004 0.051 155066 44 inner cell mass cell proliferation 5 0.004 0.053 27339 45 glutamate secretion 19 0.004 0.053 627 46 subpallium development 11 0.004 0.053 220, 585 47 ATP metabolic process 96 0.004 0.056 155066, 482 48 transcription-coupled nucleotide-excision repair 5 0.004 0.057 2073

49 striatum development 9 0.004 0.058 220, 585 50 cellular amine metabolic process 306 0.005 0.061 10157, 223, 2954, 54938, 80222 51 electron transport chain 109 0.005 0.061 4704 52 ATP synthesis coupled proton transport 38 0.005 0.061 155066 53 energy coupled proton transport, down electrochemical gradient 38 0.005 0.061 155066 54 ATP biosynthetic process 86 0.005 0.061 155066, 482 55 establishment of melanosome localization 10 0.005 0.064 585 56 RNA processing 557 0.005 0.064 23070, 26995, 27339, 5393, 56342, 6100, 6839, 85437, 9360 57 amine metabolic process 403 0.005 0.065 10157, 223, 2954, 54938, 80222 58 aerobic respiration 34 0.005 0.065 59 amine catabolic process 78 0.006 0.071 10157, 2954 60 second-messenger-mediated signaling 261 0.006 0.072 56342, 8364, 8366 61 RNA methylation 11 0.006 0.072 23070 62 oxidation reduction 614 0.006 0.073 10157, 220, 223, 242, 4704, 728294, 7923 63 nucleobase, nucleoside and nucleotide biosynthetic process 290 0.007 0.077 155066, 482, 56342, 8382 nucleobase, nucleoside, nucleotide and nucleic acid biosynthetic 64 process 290 0.007 0.077 155066, 482, 56342, 8382 65 cellular amino acid and derivative metabolic process 355 0.007 0.079 10157, 223, 2954, 54938, 80222 66 regulation of adenylate cyclase activity 96 0.007 0.079 56342 67 establishment of pigment granule localization 11 0.008 0.083 585 68 amino acid activation 45 0.008 0.083 54938, 80222 69 tRNA aminoacylation 45 0.008 0.083 54938, 80222 70 tRNA aminoacylation for protein translation 45 0.008 0.083 54938, 80222 71 dicarboxylic acid metabolic process 42 0.008 0.084

157 72 flagellum assembly 7 0.008 0.084 585 73 flagellum organization 7 0.008 0.084 585

74 small molecule catabolic process 419 0.008 0.084 10157, 254272, 2954, 79077 75 purine nucleotide biosynthetic process 246 0.008 0.084 155066, 482, 56342, 8382 76 cognition 115 0.008 0.084 585, 627, 6529 77 nucleoside triphosphate metabolic process 244 0.008 0.085 155066, 254272, 482, 79077, 8382 78 GPI anchor biosynthetic process 33 0.009 0.087 79087, 80235 79 purine nucleotide metabolic process 422 0.009 0.088 155066, 254272, 482, 56342, 8382 80 RNA modification 49 0.009 0.088 23070, 26995 81 melanosome localization 14 0.009 0.088 585 82 regulation of cAMP biosynthetic process 111 0.009 0.089 56342 83 vesicle docking involved in exocytosis 25 0.009 0.091 84 regulation of cytokinesis 9 0.009 0.091 585 85 cerebellar Purkinje cell differentiation 6 0.009 0.091 86 cerebellar Purkinje cell layer formation 6 0.009 0.091 87 cerebellar Purkinje cell layer morphogenesis 6 0.009 0.091 88 GPI anchor metabolic process 34 0.01 0.093 79087, 80235 89 regulation of lyase activity 99 0.01 0.095 56342 90 cAMP biosynthetic process 113 0.01 0.096 56342 91 regulation of cyclase activity 98 0.01 0.099 56342 92 cerebellar cortex development 14 0.01 0.099

Supplemental Table 4.3. Among genes with higher expression in Alzheimer’s, several biological processes were enriched.

Rank Name #Genes P-Value FDR SigGenes 1 translational elongation 101 4.07E-20 1.69E-16 11224, 1937, 2197, 25873, 6135, 6154, 6173, 6176, 6188, 6193, 6207, 6222, 6232 10219, 10410, 1050, 10581, 11326, 12, 23643, 241, 2532, 2919, 313, 3440, 3487, 3588, 3600, 3823, 4057, 4615, 4688, 51191, 2 defense response 661 1.16E-16 2.42E-13 58191, 6039, 604, 60675, 6283, 7097, 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 7305, 8519, 929, 9332, 9450 10346, 10410, 10581, 11326, 23643, 2669, 2919, 3108, 3109, 3122, 3588, 3600, 4057, 4615, 4688, 4860, 51191, 54209, 563, 3 immune response 660 3.86E-15 5.34E-12 58191, 604, 6398, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719, 8519, 929, 9450 10219, 1050, 11326, 12, 1462, 2162, 23643, 241, 2697, 2919, 313, 3399, 3440, 3487, 3587, 3588, 3600, 4615, 4814, 604, 4 response to wounding 598 1.07E-14 1.11E-11 60675, 6283, 7097, 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 7423, 929, 9332, 9450 5 innate immune response 201 8.28E-13 6.68E-10 10410, 10581, 11326, 23643, 4615, 4688, 51191, 58191, 7097, 7098, 710, 7100, 712, 713, 714, 718, 8519, 929, 9450 10219, 1050, 11326, 12, 23643, 241, 2919, 313, 3440, 3487, 3588, 3600, 4615, 604, 60675, 6283, 7097, 7098, 710, 7100, 712, 6 inflammatory response 365 9.65E-13 6.68E-10 713, 7132, 714, 718, 719, 929, 9332, 9450 10049, 10346, 10410, 10581, 11080, 1373, 1937, 23643, 3315, 3440, 3587, 3600, 3665, 3669, 4057, 51191, 6283, 7079, 7097, 7 response to biotic stimulus 442 8.34E-12 3.85E-09 7098, 7100, 7132, 8519, 871, 929 10346, 10410, 10581, 1373, 1937, 23643, 3315, 3440, 3587, 3600, 3665, 3669, 4057, 51191, 6283, 7079, 7097, 7098, 7100, 8 response to other organism 354 5.69E-11 2.36E-08 7132, 8519, 929 9 activation of immune response 110 7.45E-11 2.81E-08 11326, 1997, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719 10 regulation of immune response 245 8.89E-10 3.08E-07 11326, 1997, 3600, 604, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719 11 response to bacterium 213 1.42E-09 4.53E-07 1373, 23643, 3587, 4057, 6283, 7079, 7097, 7098, 7100, 7132, 929 12 I-kappaB kinase/NF-kappaB cascade 171 2.35E-09 6.98E-07 23643, 2697, 3965, 4615, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132 158 13 regulation of toll-like receptor signaling pathway 9 2.97E-09 8.21E-07 7097, 7098, 7100 14 positive regulation of toll-like receptor signaling pathway 7 7.11E-09 1.85E-06 7097, 7098, 7100

15 humoral immune response 80 8.56E-09 2.09E-06 11326, 4057, 54209, 710, 712, 713, 714, 718, 9450 16 positive regulation of immune response 155 9.07E-09 2.09E-06 11326, 1997, 3600, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719 17 positive regulation of intracellular protein kinase cascade 222 1.03E-08 2.24E-06 2697, 3059, 3965, 4615, 6275, 6283, 6398, 7098, 7105, 7132, 7423 18 regulation of response to stimulus 533 1.94E-08 4.02E-06 10488, 11326, 1997, 23411, 285, 2874, 3600, 4615, 604, 6188, 7097, 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 7423 19 positive regulation of immune system process 268 2.77E-08 5.36E-06 11326, 1997, 3600, 4860, 604, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719, 7423 20 defense response to bacterium 96 2.84E-08 5.36E-06 4057, 6283, 7097, 7098, 7100, 7132 positive regulation of I-kappaB kinase/NF-kappaB 21 cascade 115 4.25E-08 7.67E-06 2697, 3965, 4615, 6275, 6283, 6398, 7098, 7105, 7132 humoral immune response mediated by circulating 22 immunoglobulin 37 4.47E-08 7.73E-06 710, 712, 713, 714, 718 23 positive regulation of response to stimulus 276 5.08E-08 8.43E-06 10488, 11326, 1997, 3600, 7097, 7098, 710, 7100, 712, 713, 7132, 714, 718, 719, 7423 24 complement activation 37 7.95E-08 1.24E-05 11326, 710, 712, 713, 714, 718 25 toll-like receptor signaling pathway 22 8.06E-08 1.24E-05 7097, 7098, 7100 26 regulation of I-kappaB kinase/NF-kappaB cascade 127 8.46E-08 1.25E-05 2697, 3965, 4615, 6275, 6283, 6398, 7098, 7105, 7132 27 regulation of cytokine production 199 9.39E-08 1.34E-05 11326, 1997, 4615, 604, 7097, 7098, 7100, 718, 719, 929 28 innate immune response-activating signal transduction 25 1.02E-07 1.41E-05 7097, 7098, 7100 29 pattern recognition receptor signaling pathway 24 1.11E-07 1.49E-05 7097, 7098, 7100 30 activation of innate immune response 26 1.24E-07 1.61E-05 7097, 7098, 7100 31 cytokine production 226 2.25E-07 2.83E-05 11326, 1997, 4615, 4860, 604, 7097, 7098, 7100, 718, 719, 929 32 positive regulation of JNK cascade 23 2.34E-07 2.86E-05 7098 33 positive regulation of signal transduction 275 3.05E-07 3.58E-05 2697, 3059, 3965, 4615, 4734, 6275, 6283, 6398, 7098, 7105, 7132, 7423 34 positive regulation of tumor necrosis factor production 14 3.11E-07 3.58E-05 7097, 7098, 929 35 positive regulation of signaling process 280 3.59E-07 3.97E-05 2697, 3059, 3965, 4615, 4734, 6275, 6283, 6398, 7098, 7105, 7132, 7423 36 interleukin-6 production 44 3.63E-07 3.97E-05 4615, 7097, 7098 37 complement activation, classical pathway 27 4.92E-07 5.24E-05 710, 712, 713, 714, 718 38 positive regulation of cytokine production 104 5.08E-07 5.28E-05 4615, 7097, 7098, 7100, 718, 719, 929 39 positive regulation of interleukin-6 production 23 7.62E-07 7.54E-05 4615, 7097, 7098 40 regulation of interleukin-12 production 23 9.39E-07 9.07E-05 7097, 7098 41 positive regulation of stress-activated protein kinase 29 1.08E-06 1.02E-04 7098

signaling cascade 10346, 10410, 10488, 1050, 10581, 1373, 1937, 1981, 200186, 23411, 23643, 23764, 285, 3315, 3440, 358, 3587, 3600, 3665, 42 multi-organism process 772 1.17E-06 1.08E-04 3669, 4057, 4734, 51191, 5422, 5696, 6283, 6993, 7079, 7097, 7098, 7100, 7132, 8519, 929 43 interleukin-12 production 24 1.31E-06 1.15E-04 7097, 7098 44 positive regulation of signaling pathway 376 1.35E-06 1.15E-04 1950, 2697, 3059, 3965, 4615, 4734, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132, 7423 45 positive regulation of cell communication 408 1.64E-06 1.15E-04 1950, 2697, 3059, 3965, 4615, 4734, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132, 7423 46 response to molecule of bacterial origin 133 1.78E-06 1.15E-04 1373, 23643, 3587, 7079, 7097, 7100, 7132, 929 47 regulation of response to stress 321 1.87E-06 1.15E-04 23411, 2874, 3600, 4615, 604, 6188, 7097, 7098, 710, 7100, 7132, 718, 7423 modulation by organism of defense response of other 48 organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 modulation by organism of immune response of other 49 organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 modulation by organism of innate immunity in other 50 organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 51 modulation by symbiont of host defense response 6 1.90E-06 1.15E-04 7097, 7098, 7100 52 modulation by symbiont of host immune response 6 1.90E-06 1.15E-04 7097, 7098, 7100 53 modulation by symbiont of host innate immunity 6 1.90E-06 1.15E-04 7097, 7098, 7100 pathogen-associated molecular pattern dependent induction by organism of innate immunity of other 54 organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 pathogen-associated molecular pattern dependent 55 induction by symbiont of host innate immunity 6 1.90E-06 1.15E-04 7097, 7098, 7100 pathogen-associated molecular pattern dependent modulation by organism of innate immunity in other 56 organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100

159 pathogen-associated molecular pattern dependent 57 modulation by symbiont of host innate immunity 6 1.90E-06 1.15E-04 7097, 7098, 7100

positive regulation by organism of defense response of 58 other organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 positive regulation by organism of immune response of 59 other organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 positive regulation by organism of innate immunity in 60 other organism involved in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 61 positive regulation by symbiont of host defense response 6 1.90E-06 1.15E-04 7097, 7098, 7100 62 positive regulation by symbiont of host immune response 6 1.90E-06 1.15E-04 7097, 7098, 7100 63 positive regulation by symbiont of host innate immunity 6 1.90E-06 1.15E-04 7097, 7098, 7100 64 response to host immune response 6 1.90E-06 1.15E-04 7097, 7098, 7100 response to immune response of other organism involved 65 in symbiotic interaction 6 1.90E-06 1.15E-04 7097, 7098, 7100 66 immune effector process 210 1.91E-06 1.15E-04 11326, 3600, 604, 7097, 7098, 710, 712, 713, 714, 718 67 immunoglobulin mediated immune response 70 2.99E-06 1.77E-04 604, 710, 712, 713, 714, 718 68 B cell mediated immunity 71 3.90E-06 2.25E-04 604, 710, 712, 713, 714, 718 activation of plasma proteins involved in acute 69 inflammatory response 40 4.94E-06 2.81E-04 11326, 710, 712, 713, 714, 718 70 regulation of interleukin-6 production 42 5.19E-06 2.91E-04 4615, 7097, 7098 71 positive regulation of interleukin-8 production 10 5.38E-06 2.98E-04 7097, 7098, 7100 72 regulation of transcription regulator activity 146 5.78E-06 3.16E-04 23411, 3399, 6188, 7097, 7098 73 regulation of immune system process 424 5.94E-06 3.20E-04 11326, 1997, 3600, 4860, 604, 7097, 7098, 710, 7100, 712, 713, 714, 718, 719, 7423, 8399 74 macrophage activation involved in immune response 10 6.04E-06 3.22E-04 7097, 7098, 7100 75 regulation of transcription factor activity 145 6.66E-06 3.46E-04 23411, 3399, 6188, 7097, 7098 antigen processing and presentation of peptide or 76 polysaccharide antigen via MHC class II 13 6.67E-06 3.46E-04 3108, 3109, 3122 modification by symbiont of host morphology or 77 physiology 9 6.88E-06 3.53E-04 7097, 7098, 7100 modification of morphology or physiology of other 78 organism involved in symbiotic interaction 21 7.22E-06 3.65E-04 358, 7097, 7098, 7100

79 regulation of innate immune response 61 7.69E-06 3.85E-04 7097, 7098, 710, 7100 regulation of vascular endothelial growth factor receptor 80 signaling pathway 16 1.09E-05 5.30E-04 4734, 7423 81 positive regulation of angiogenesis 43 1.15E-05 5.54E-04 285, 358, 7132, 718, 719, 7423 82 symbiosis, encompassing mutualism through parasitism 64 1.17E-05 5.57E-04 358, 4734, 6993, 7097, 7098, 7100 83 positive regulation of type I interferon production 15 1.18E-05 5.58E-04 7097, 7098 84 microglial cell activation involved in immune response 7 1.25E-05 5.81E-04 7097, 7098, 7100 85 positive regulation of interleukin-12 production 12 1.26E-05 5.81E-04 7097, 7098 86 regulation of defense response 156 1.29E-05 5.90E-04 3600, 4615, 604, 7097, 7098, 710, 7100, 7132, 718 87 cellular response to UV 6 1.31E-05 5.92E-04 358, 4734 adaptive immune response based on somatic recombination of immune receptors built from 88 immunoglobulin superfamily domains 114 1.47E-05 6.58E-04 604, 710, 712, 713, 714, 718 89 positive regulation of multicellular organismal process 279 1.53E-05 6.76E-04 1373, 358, 4615, 60675, 7097, 7098, 7100, 718, 719, 929, 9388 90 cytokine production involved in immune response 22 1.60E-05 6.82E-04 604, 7097 response to defenses of other organism involved in 91 symbiotic interaction 8 1.62E-05 6.82E-04 7097, 7098, 7100 92 response to host 8 1.62E-05 6.82E-04 7097, 7098, 7100 93 response to host defenses 8 1.62E-05 6.82E-04 7097, 7098, 7100 94 adaptive immune response 116 1.63E-05 6.82E-04 604, 710, 712, 713, 714, 718 95 antigen processing and presentation 50 1.69E-05 7.04E-04 3108, 3109, 3122, 563 96 translation 394 2.07E-05 8.51E-04 11224, 1937, 1978, 1981, 2197, 25873, 283, 3315, 6135, 6154, 6173, 6176, 6188, 6193, 6207, 6222, 6232, 92399 vascular endothelial growth factor receptor signaling 97 pathway 22 2.21E-05 8.99E-04 4734, 7423

160 98 regulation of intracellular protein kinase cascade 328 2.33E-05 9.38E-04 2697, 2874, 3059, 3965, 4615, 6275, 6283, 6398, 7098, 7105, 7132, 7423 99 regulation of type I interferon production 19 2.57E-05 1.03E-03 7097, 7098 100 immune response-activating signal transduction 72 2.62E-05 1.04E-03 1997, 7097, 7098, 7100, 719

101 acute inflammatory response 100 2.87E-05 1.10E-03 1050, 11326, 12, 710, 712, 713, 714, 718, 9332 102 cellular response to light stimulus 7 2.87E-05 1.10E-03 358, 4734 103 immune response-regulating signaling pathway 80 3.19E-05 1.20E-03 1997, 7097, 7098, 7100, 719 104 detection of biotic stimulus 20 3.38E-05 1.27E-03 23643, 7097, 7098 105 type I interferon production 21 3.84E-05 1.42E-03 7097, 7098 regulation of cytokine production involved in immune 106 response 19 3.96E-05 1.46E-03 604, 7097 107 interferon-beta production 17 4.07E-05 1.47E-03 7097, 7098 108 regulation of interferon-beta production 17 4.07E-05 1.47E-03 7097, 7098 109 positive regulation of interferon-beta production 13 4.10E-05 1.47E-03 7097, 7098 110 I-kappaB phosphorylation 11 5.28E-05 1.86E-03 7097, 7098 111 positive regulation of MAPKKK cascade 87 8.00E-05 0.003 7098, 7423 112 anti-apoptosis 228 8.38E-05 0.003 1410, 23411, 25816, 26574, 3315, 4615, 57099, 60675, 7423, 81542, 8870, 9531 113 regulation of humoral immune response 13 8.59E-05 0.003 710, 718 114 negative regulation of cytokine production 42 8.86E-05 0.003 11326, 604, 7097 115 intracellular protein kinase cascade 583 8.90E-05 0.003 1410, 1950, 23643, 2697, 2874, 3059, 3965, 4615, 60675, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132, 7423, 9459 116 signal transmission via phosphorylation event 583 8.90E-05 0.003 1410, 1950, 23643, 2697, 2874, 3059, 3965, 4615, 60675, 6275, 6283, 6398, 7097, 7098, 7100, 7105, 7132, 7423, 9459 117 positive regulation of NF-kappaB import into nucleus 15 9.03E-05 0.003 7097, 7098 118 regulation of immune effector process 103 9.09E-05 0.003 3600, 604, 7097, 710, 718 119 negative regulation of transcription factor activity 61 9.53E-05 0.003 23411, 3399, 6188 120 negative regulation of transcription regulator activity 61 9.53E-05 0.003 23411, 3399, 6188 121 regulation of angiogenesis 87 9.73E-05 0.003 285, 358, 7132, 718, 719, 7423 122 interferon-beta biosynthetic process 7 1.01E-04 0.003 7098 123 positive regulation of interferon-beta biosynthetic process 7 1.01E-04 0.003 7098 124 regulation of interferon-beta biosynthetic process 7 1.01E-04 0.003 7098 125 regulation of lipid transport 40 1.05E-04 0.003 706, 8399, 9388 126 lymphocyte mediated immunity 108 1.06E-04 0.003 604, 710, 712, 713, 714, 718 127 cellular response to radiation 9 1.07E-04 0.003 358, 4734

128 type I interferon biosynthetic process 10 1.12E-04 0.003 7098 129 tumor necrosis factor superfamily cytokine production 36 1.19E-04 0.003 7097, 7098, 929 130 regulation of DNA binding 165 1.19E-04 0.003 23411, 3399, 6188, 7097, 7098 131 blood vessel morphogenesis 276 1.30E-04 0.004 1950, 2697, 283, 285, 358, 60675, 7132, 718, 719, 7423 132 positive regulation of protein transport 83 1.43E-04 0.004 1950, 283, 7097, 7098 133 positive regulation of innate immune response 50 1.48E-04 0.004 7097, 7098, 7100 134 negative regulation of immune effector process 18 1.57E-04 0.004 604, 710 regulation of production of molecular mediator of immune 135 response 41 2.06E-04 0.006 604, 7097 136 response to lipopolysaccharide 122 2.14E-04 0.006 1373, 23643, 3587, 7079, 7097, 7132, 929 137 leukocyte mediated immunity 133 2.22E-04 0.006 604, 710, 712, 713, 714, 718 138 myeloid leukocyte cytokine production 9 2.23E-04 0.006 604, 7097 139 translational initiation 60 2.34E-04 0.006 1978, 1981, 3315, 6188, 6193 140 negative regulation of DNA binding 69 2.44E-04 0.006 23411, 3399, 6188 141 angiogenesis 234 2.45E-04 0.006 1950, 283, 285, 358, 60675, 7132, 718, 719, 7423 142 regulation of tumor necrosis factor production 33 2.57E-04 0.007 7097, 7098, 929 143 tumor necrosis factor production 33 2.57E-04 0.007 7097, 7098, 929 144 response to mercury ion 9 2.60E-04 0.007 358, 5224 145 positive regulation of intracellular transport 46 2.64E-04 0.007 1950, 4734, 7097, 7098 146 prostaglandin metabolic process 25 2.73E-04 0.007 27306, 6916, 7132 147 prostanoid metabolic process 25 2.73E-04 0.007 27306, 6916, 7132 148 response to exogenous dsRNA 11 2.82E-04 0.007 7098 149 wound healing 215 3.28E-04 0.008 2162, 2697, 3587, 3588, 4814, 710, 7423 150 response to virus 153 3.49E-04 0.009 10346, 10410, 10581, 1937, 3315, 3440, 3600, 3665, 3669, 51191, 7098, 8519

161 151 defense response to Gram-positive bacterium 29 3.51E-04 0.009 7097 152 regulation of binding 209 3.55E-04 0.009 23411, 3399, 6188, 7097, 7098

153 notochord development 7 3.92E-04 0.009 3399 10049, 10076, 1050, 11080, 1373, 1410, 1978, 2052, 23643, 2697, 283, 285, 3059, 3315, 3399, 358, 3587, 4615, 58191, 6402, 154 response to organic substance 886 4.14E-04 0.01 7079, 7097, 7098, 7100, 713, 7132, 7913, 871, 900, 929 155 positive regulation of phosphorylation 135 4.28E-04 0.01 1950, 283, 3059, 7423 156 regulation of gene-specific transcription 220 4.29E-04 0.01 1050, 23411, 2874, 7097, 7098 157 detection of molecule of bacterial origin 8 4.72E-04 0.011 23643, 7097 regulation of stress-activated protein kinase signaling 158 cascade 90 4.74E-04 0.011 2874, 7098 159 NF-kappaB import into nucleus 26 4.81E-04 0.011 7097, 7098 160 regulation of NF-kappaB import into nucleus 26 4.81E-04 0.011 7097, 7098 161 positive regulation of defense response 85 5.03E-04 0.012 7097, 7098, 7100, 7132, 718 162 response to protein stimulus 115 5.26E-04 0.012 10049, 11080, 285, 3315, 3399, 7913, 871 163 negative regulation of binding 84 5.48E-04 0.012 23411, 3399, 6188 164 positive regulation of transmembrane transport 26 5.90E-04 0.013 7097, 7098 10076, 10488, 11037, 1410, 1950, 23411, 2697, 283, 285, 358, 4734, 58191, 5867, 604, 706, 7097, 7098, 718, 719, 7423, 8399, 165 regulation of localization 726 6.27E-04 0.014 84876, 91624, 9388 166 prostaglandin biosynthetic process 16 6.29E-04 0.014 27306, 6916 167 prostanoid biosynthetic process 16 6.29E-04 0.014 27306, 6916 168 protein maturation 116 6.30E-04 0.014 11326, 710, 712, 713, 714, 718, 871, 9049 169 chemokine production 26 6.32E-04 0.014 7097, 7098 170 regulation of cell fate commitment 9 6.61E-04 0.014 22943 171 regulation of cell fate specification 9 6.61E-04 0.014 22943 172 regulation of cellular response to stress 138 6.76E-04 0.015 23411, 2874, 6188, 7098 173 interferon-alpha production 9 7.27E-04 0.015 7098 174 regulation of interferon-alpha production 9 7.27E-04 0.015 7098 175 positive regulation of chemokine production 16 7.36E-04 0.016 7097, 7098 176 myeloid cell activation involved in immune response 27 7.80E-04 0.016 7097, 7098, 7100 177 cellular response to molecule of bacterial origin 28 8.45E-04 0.018 7097, 929 178 stress-activated protein kinase signaling cascade 122 9.49E-04 0.02 1410, 2874, 7098, 9459

179 blood vessel development 324 9.91E-04 0.02 1950, 2697, 283, 285, 358, 60675, 7132, 718, 719, 7423 180 regeneration 86 9.98E-04 0.021 10076, 1050, 1462, 2697, 285, 4814 positive regulation of transcription factor import into 181 nucleus 20 1.05E-03 0.021 7097, 7098 182 regulation of JNK cascade 81 1.06E-03 0.022 2874, 7098 183 negative regulation of smooth muscle cell proliferation 13 1.08E-03 0.022 283, 3600 184 positive regulation of transport 276 1.12E-03 0.023 10488, 1950, 283, 358, 4734, 7097, 7098, 718, 8399, 84876, 9388 185 regulation of transport 514 1.17E-03 0.023 10488, 11037, 1410, 1950, 23411, 2697, 283, 358, 4734, 5867, 706, 7097, 7098, 718, 8399, 84876, 9388 186 response to pH 13 1.17E-03 0.023 2697, 84329 gene-specific transcription from RNA polymerase II 187 promoter 161 1.21E-03 0.024 1050, 23411, 2874, 7097, 7098 regulation of gene-specific transcription from RNA 188 polymerase II promoter 161 1.21E-03 0.024 1050, 23411, 2874, 7097, 7098 189 positive regulation of cytokine biosynthetic process 53 1.35E-03 0.026 7097, 7098 190 icosanoid metabolic process 51 1.35E-03 0.026 241, 27306, 6916, 7132, 8399 191 positive regulation of phosphate metabolic process 138 1.44E-03 0.028 1950, 283, 3059, 7423 192 positive regulation of phosphorus metabolic process 138 1.44E-03 0.028 1950, 283, 3059, 7423 193 negative regulation of immune system process 90 1.45E-03 0.028 11326, 1997, 604, 710 194 regulation of interferon-gamma production 34 1.51E-03 0.028 7098 195 microglial cell activation 10 1.54E-03 0.029 7097, 7098, 7100 196 positive regulation of chemokine biosynthetic process 8 1.59E-03 0.029 7098 197 positive regulation of protein import into nucleus 25 1.64E-03 0.03 7097, 7098 198 response to unfolded protein 65 1.68E-03 0.03 10049, 11080, 3315, 871 199 chemokine biosynthetic process 12 1.70E-03 0.03 7098 162 200 chemokine metabolic process 12 1.70E-03 0.03 7098 201 regulation of B cell apoptosis 8 1.70E-03 0.03 604

202 negative regulation of cell death 402 1.71E-03 0.03 1410, 23411, 25816, 26574, 3315, 358, 4615, 57099, 604, 60675, 7423, 81542, 8870, 900, 9531 203 cellular response to xenobiotic stimulus 27 1.71E-03 0.03 6283, 9049 204 xenobiotic metabolic process 27 1.71E-03 0.03 6283, 9049 205 interferon-gamma production 35 1.75E-03 0.031 7098 206 response to xenobiotic stimulus 31 1.80E-03 0.032 6283, 9049 207 cellular defense response 51 1.80E-03 0.032 10219, 23643, 3823, 4688, 7305 208 response to fungus 22 1.82E-03 0.032 6283, 7097 209 negative regulation of apoptosis 391 1.83E-03 0.032 1410, 23411, 25816, 26574, 3315, 358, 4615, 57099, 604, 60675, 7423, 81542, 8870, 900, 9531 210 S phase of mitotic cell cycle 25 1.85E-03 0.032 5422, 5424, 604, 8099, 84967 211 regulation of mononuclear cell proliferation 93 1.85E-03 0.032 11326, 3600, 4860, 604 212 regulation of lymphocyte proliferation 92 1.88E-03 0.032 11326, 3600, 4860, 604 213 positive regulation of nucleocytoplasmic transport 32 1.92E-03 0.033 4734, 7097, 7098 214 negative regulation of programmed cell death 396 1.93E-03 0.033 1410, 23411, 25816, 26574, 3315, 358, 4615, 57099, 604, 60675, 7423, 81542, 8870, 900, 9531 215 regulation of leukocyte proliferation 94 1.95E-03 0.033 11326, 3600, 4860, 604 216 negative regulation of lipid transport 15 1.96E-03 0.033 217 negative regulation of multi-organism process 18 1.96E-03 0.033 7097 218 cell migration 455 1.97E-03 0.033 10076, 10488, 128954, 1462, 2191, 2697, 283, 285, 4478, 58191, 7097, 719, 7423, 91624 219 regulation of digestive system process 14 1.99E-03 0.033 358 220 regulation of G2/M transition of mitotic cell cycle 9 2.00E-03 0.033 900 221 cellular response to abiotic stimulus 13 2.02E-03 0.033 358, 4734 222 JNK cascade 109 2.03E-03 0.033 2874, 7098, 9459 223 macrophage activation 27 2.05E-03 0.034 7097, 7098, 7100, 8399 224 response to peptidoglycan 11 2.09E-03 0.034 7097 225 unsaturated fatty acid metabolic process 54 2.16E-03 0.035 241, 27306, 6916, 7132, 8399 226 detection of external stimulus 71 2.17E-03 0.035 56925, 7097, 7098 227 lymphocyte proliferation 122 2.40E-03 0.038 11326, 3600, 4860, 604 228 mononuclear cell proliferation 124 2.42E-03 0.038 11326, 3600, 4860, 604 229 regulation of cholesterol transport 23 2.43E-03 0.038 706, 9388 230 regulation of sterol transport 23 2.43E-03 0.038 706, 9388

231 negative regulation of response to stimulus 122 2.47E-03 0.038 23411, 285, 2874, 604, 6188, 710 232 vasculature development 334 2.47E-03 0.038 1950, 2697, 283, 285, 358, 60675, 7132, 718, 719, 7423 233 interaction with host 39 0.003 0.039 4734, 6993, 7097, 7098, 7100 234 cytokine biosynthetic process 89 0.003 0.04 7097, 7098 235 cytokine metabolic process 90 0.003 0.04 7097, 7098 236 response to glucocorticoid stimulus 97 0.003 0.04 10076, 1050, 1373, 358, 713 237 cell activation involved in immune response 55 0.003 0.041 604, 7097, 7098, 7100 238 leukocyte activation involved in immune response 55 0.003 0.041 604, 7097, 7098, 7100 239 locomotion 599 0.003 0.041 10076, 10488, 128954, 1462, 2191, 2697, 283, 285, 2919, 4478, 4734, 5355, 58191, 6036, 60675, 7097, 719, 7423, 91624 positive regulation of production of molecular mediator of 240 immune response 15 0.003 0.041 7097 241 regulation of transcription in response to stress 6 0.003 0.041 4734 regulation of transcription from RNA polymerase II 242 promoter 745 0.003 0.043 100125288, 1050, 1997, 23411, 26574, 2874, 29128, 3399, 3660, 3665, 4734, 57658, 5932, 604, 7097, 7098, 7132, 7913, 79366 243 positive regulation of lipid transport 20 0.003 0.043 8399, 9388 244 regulation of protein secretion 63 0.003 0.043 1950, 283 10488, 1050, 1373, 1509, 23411, 285, 2919, 358, 4615, 5355, 56925, 58191, 6036, 604, 60675, 7097, 7098, 710, 7132, 718, 245 response to external stimulus 645 0.003 0.043 719, 7423, 9388 246 leukocyte proliferation 126 0.003 0.043 11326, 3600, 4860, 604 247 leukocyte activation 336 0.003 0.043 11326, 3600, 4860, 604, 7097, 7098, 7100, 81542, 8399 248 ribosomal large subunit biogenesis 11 0.003 0.043 6135, 6154 249 cellular response to lipoteichoic acid 6 0.003 0.043 7097, 929 250 response to lipoteichoic acid 6 0.003 0.043 7097, 929 251 positive regulation of intracellular protein transport 43 0.003 0.047 1950, 7097, 7098 163 252 positive regulation of ERK1 and ERK2 cascade 34 0.003 0.047 7423 253 lymphocyte chemotaxis 9 0.003 0.047 58191

254 protein maturation by peptide bond cleavage 81 0.003 0.047 11326, 710, 712, 713, 714, 718 255 response to corticosteroid stimulus 104 0.003 0.047 10076, 1050, 1373, 358, 713 256 response to dsRNA 27 0.003 0.048 7098 257 peptidyl-tyrosine modification 112 0.003 0.049 1950, 3059, 7423 258 RNA catabolic process 74 0.004 0.049 1763, 3669, 6036, 6039, 87178 259 tetrahydrobiopterin biosynthetic process 6 0.004 0.049 6697 260 cellular response to inorganic substance 11 0.004 0.05 241, 358 261 cellular response to metal ion 11 0.004 0.05 241, 358 1410, 1462, 1509, 1763, 23411, 26574, 27244, 2874, 29128, 358, 3978, 4438, 4734, 5422, 5424, 5932, 604, 6188, 7098, 79677, 262 cellular response to stress 677 0.004 0.05 83932, 900, 9459 263 positive regulation of transcription regulator activity 88 0.004 0.05 7097, 7098 positive regulation of NF-kappaB transcription factor 264 activity 58 0.004 0.051 7097, 7098 265 negative regulation of lymphocyte apoptosis 9 0.004 0.051 604 266 response to heat 62 0.004 0.052 11080, 1410, 3315, 929 267 regulation of epidermis development 21 0.004 0.053 23764 268 regulation of protein transport 145 0.004 0.053 1950, 23411, 283, 7097, 7098 269 positive regulation of cellular component movement 138 0.004 0.053 10488, 58191, 604, 7097, 719, 7423 270 positive regulation of transcription factor activity 87 0.004 0.055 7097, 7098 271 establishment of mitotic spindle orientation 7 0.004 0.057 6993 272 establishment of spindle orientation 7 0.004 0.057 6993 273 cell activation 381 0.005 0.059 11326, 3600, 4860, 604, 7097, 7098, 7100, 81542, 8399 274 response to abiotic stimulus 393 0.005 0.06 11080, 1410, 2697, 2766, 285, 3315, 358, 4734, 5424, 56925, 7098, 84329, 929 275 receptor-mediated endocytosis 73 0.005 0.061 10268, 4734, 58191 276 regulation of inflammatory response 87 0.005 0.061 4615, 604, 7097, 7098, 710, 7132, 718 277 pancreatic juice secretion 8 0.005 0.061 358 278 mitotic cell cycle 480 0.005 0.061 10459, 1950, 1978, 246184, 26574, 5422, 5424, 54443, 54930, 55145, 5696, 604, 6993, 8099, 81930, 84967, 900, 90293 279 cell cycle phase 509 0.005 0.061 10459, 1950, 1978, 246184, 4438, 5422, 5424, 54443, 54930, 604, 6993, 8099, 81930, 84967, 900, 90293 280 regulation of cytokine biosynthetic process 79 0.005 0.062 7097, 7098

281 protein processing 106 0.005 0.064 11326, 710, 712, 713, 714, 718 282 regulation of gastrulation 9 0.005 0.067 positive regulation of cytokine production involved in 283 immune response 11 0.005 0.067 7097 284 regulation of lymphocyte apoptosis 15 0.006 0.067 604 285 regulation of interleukin-8 production 21 0.006 0.067 7097, 7098, 7100 286 regulation of response to external stimulus 187 0.006 0.069 10488, 285, 4615, 604, 7097, 7098, 710, 7132, 718, 719, 7423 287 interferon-gamma biosynthetic process 14 0.006 0.069 7098 288 peptidyl-tyrosine phosphorylation 110 0.006 0.069 1950, 3059, 7423 289 complement activation, alternative pathway 14 0.006 0.069 11326, 718 290 regulation of cell activation 189 0.006 0.07 11326, 3600, 4860, 604, 8399 positive regulation of interferon-gamma biosynthetic 291 process 11 0.006 0.07 7098 292 positive regulation of protein secretion 45 0.006 0.071 283 293 regulation of multi-organism process 52 0.006 0.072 3600, 7097 294 production of molecular mediator of immune response 64 0.006 0.072 604, 7097 295 interphase of mitotic cell cycle 137 0.006 0.072 1978, 5422, 5424, 604, 8099, 84967, 900 296 rRNA transcription 18 0.006 0.072 283 297 positive regulation of humoral immune response 8 0.006 0.072 718 100125288, 1050, 1997, 23411, 23764, 26574, 2874, 29128, 29777, 3399, 3660, 3665, 4734, 57658, 5932, 604, 7097, 7098, 298 transcription from RNA polymerase II promoter 900 0.006 0.073 7132, 7913, 79366 299 interleukin-8 production 23 0.007 0.075 7097, 7098, 7100 300 cell cycle process 662 0.007 0.076 10459, 1950, 1978, 246184, 27244, 4438, 5422, 5424, 54443, 54930, 5696, 604, 6993, 8099, 81930, 84967, 900, 90293 301 regulation of establishment of protein localization 153 0.007 0.076 1950, 23411, 283, 7097, 7098 164 302 mesenchyme development 81 0.007 0.077 1592, 6275 303 regulation of leukocyte migration 40 0.007 0.077 7097, 719, 7423

304 myeloid leukocyte activation 71 0.007 0.077 7097, 7098, 7100, 8399 305 superoxide anion generation 14 0.007 0.079 26574, 4688 306 regulation of chemokine production 24 0.007 0.081 7097, 7098 307 regulation of leukocyte mediated immunity 59 0.007 0.082 604, 718 308 tetrahydrobiopterin metabolic process 7 0.007 0.083 6697 309 acute-phase response 44 0.008 0.083 1050, 12, 9332 310 regulation of chemokine biosynthetic process 11 0.008 0.083 7098 311 regulation of leukocyte activation 177 0.008 0.083 11326, 3600, 4860, 604, 8399 312 regulation of B cell mediated immunity 24 0.008 0.083 604, 718 313 regulation of immunoglobulin mediated immune response 24 0.008 0.083 604, 718 314 protein secretion 94 0.008 0.084 1950, 283, 4860 315 positive regulation of cell migration 129 0.008 0.084 10488, 58191, 7097, 719, 7423 316 defense response to virus 41 0.008 0.084 3600, 7098 317 tissue regeneration 30 0.008 0.084 2697, 4814 318 regulation of body fluid levels 200 0.008 0.084 187, 2162, 358, 3587, 3588, 710 319 positive regulation of leukocyte migration 31 0.008 0.084 7097, 719, 7423 320 positive regulation of inflammatory response 43 0.008 0.084 7097, 7098, 7132, 718 321 interspecies interaction between organisms 335 0.008 0.085 10346, 10488, 1050, 1981, 200186, 23411, 358, 3665, 4734, 5422, 5696, 6993, 7097, 7098, 7100, 7132 322 cell motility 484 0.008 0.085 10076, 10488, 128954, 1462, 2191, 2697, 283, 285, 4478, 58191, 7097, 719, 7423, 91624 323 localization of cell 484 0.008 0.085 10076, 10488, 128954, 1462, 2191, 2697, 283, 285, 4478, 58191, 7097, 719, 7423, 91624 324 cellular response to lipopolysaccharide 25 0.008 0.085 929 325 regulation of nucleocytoplasmic transport 84 0.008 0.085 1950, 23411, 4734, 7097, 7098 326 positive regulation of mononuclear cell proliferation 65 0.008 0.086 3600, 4860, 604 327 icosanoid biosynthetic process 36 0.008 0.086 241, 27306, 6916 328 T cell proliferation 86 0.008 0.086 11326, 3600, 4860 329 regulation of protein import into nucleus 67 0.008 0.086 1950, 23411, 7097, 7098 330 positive regulation of lymphocyte proliferation 64 0.008 0.086 3600, 4860, 604 331 reproductive process 791 0.009 0.087 100125288, 10149, 10488, 23411, 23764, 283, 285, 4438, 4734, 51314, 5224, 604, 60675, 6993, 7079, 7913 332 regulation of adaptive immune response based on 51 0.009 0.088 604, 718

somatic recombination of immune receptors built from immunoglobulin superfamily domains 333 positive regulation of leukocyte proliferation 66 0.009 0.088 3600, 4860, 604 334 reproduction 793 0.009 0.089 100125288, 10149, 10488, 23411, 23764, 283, 285, 4438, 4734, 51314, 5224, 604, 60675, 6993, 7079, 7913 335 interphase 146 0.009 0.091 1978, 5422, 5424, 604, 8099, 84967, 900 336 establishment of tissue polarity 12 0.009 0.091 57216 337 positive regulation of immune effector process 46 0.009 0.091 7097, 718 338 negative regulation of RNA metabolic process 412 0.009 0.091 100125288, 10049, 1050, 23411, 2874, 3399, 3660, 3665, 4734, 604, 6207, 84232 339 transepithelial transport 10 0.01 0.093 358 340 cellular response to hypoxia 6 0.01 0.093 358 341 cellular response to oxygen levels 6 0.01 0.093 358 342 renal water transport 5 0.01 0.093 358 343 regulation of phospholipase A2 activity 6 0.01 0.094 283 10076, 10488, 1050, 11326, 138151, 1950, 2014, 23411, 2697, 27244, 283, 2919, 3059, 358, 3600, 4860, 563, 604, 6282, 648, 344 regulation of cell proliferation 834 0.01 0.095 7423, 8519 345 phospholipid catabolic process 21 0.01 0.095 9388 346 response to oxygen levels 155 0.01 0.097 1410, 283, 285, 358, 51167, 54541, 7097, 9124 347 positive regulation of gene-specific transcription 145 0.01 0.097 1050, 7097, 7098 348 B cell apoptosis 11 0.01 0.098 604

165

Supplemental Table 4.4. Among genes with lower expression in Alzheimer’s, several molecular functions were enriched.

Rank Name #Genes P-Value FDR SigGenes 1 neuropeptide hormone activity 21 5.46E-07 1.89E-04 2922, 5368, 7425 2 cation transmembrane transporter activity 525 1.09E-05 0.001 155066, 482, 6529 3 inorganic cation transmembrane transporter activity 203 5.44E-05 0.005 155066, 482, 6529 4 porin activity 6 6.50E-05 0.005 10452 5 calmodulin-dependent protein kinase activity 20 1.14E-04 0.008 6 ion transmembrane transporter activity 683 1.34E-04 0.009 10452, 155066, 482, 6529 7 MAP kinase tyrosine/serine/threonine phosphatase activity 12 1.43E-04 0.009 1846, 1848 8 MAP kinase phosphatase activity 13 1.74E-04 0.009 1846, 1848 9 monovalent inorganic cation transmembrane transporter activity 160 2.38E-04 0.012 155066, 482, 6529 10 nucleoside-triphosphate diphosphatase activity 6 4.97E-04 0.021 3704, 79077 11 hydrogen ion transmembrane transporter activity 83 0.001 0.023 155066 12 NADH dehydrogenase (quinone) activity 38 0.001 0.044 4704 13 NADH dehydrogenase (ubiquinone) activity 38 0.001 0.044 4704 14 NADH dehydrogenase activity 38 0.001 0.044 4704 10452, 11182, 155066, 21, 22, 482, 15 transmembrane transporter activity 859 0.002 0.055 6529 16 cis-trans isomerase activity 36 0.002 0.068 2954, 9360 10452, 11182, 155066, 3049, 482, 17 substrate-specific transporter activity 916 0.004 0.099 6529 18 SNARE binding 29 0.004 0.113 19 mannosyltransferase activity 15 0.005 0.12 10585, 79087, 80235 20 phosphotyrosine binding 10 0.005 0.12 53 21 oxygen transporter activity 8 0.005 0.123 3049 22 ATPase activity 330 0.005 0.123 155066, 21, 22, 482, 57680 23 RNA polymerase II carboxy-terminal domain kinase activity 13 0.006 0.127 24 sodium:potassium-exchanging ATPase activity 10 0.007 0.138 482 25 calcium channel regulator activity 13 0.007 0.138 9379 26 ATPase activity, coupled 266 0.007 0.138 155066, 21, 22, 482, 57680 27 channel regulator activity 53 0.007 0.14 57730, 9379 28 gated channel activity 296 0.008 0.141 10452 29 ATPase activity, coupled to transmembrane movement of ions 67 0.008 0.141 155066, 482 30 ATPase activity, coupled to transmembrane movement of substances 100 0.009 0.146 155066, 21, 22, 482

31 oxidoreductase activity, acting on the aldehyde or oxo group of donors 32 0.009 0.146 220, 223 32 aldehyde dehydrogenase (NAD) activity 9 0.009 0.147 220, 223 33 voltage-gated channel activity 185 0.011 0.157 10452 34 voltage-gated ion channel activity 185 0.011 0.157 10452 35 O-methyltransferase activity 12 0.011 0.157 23070 10157, 220, 223, 242, 2954, 4704, 36 oxidoreductase activity 648 0.011 0.157 728294, 7923 155066, 21, 22, 3704, 4636, 482, 37 hydrolase activity, acting on acid anhydrides 742 0.012 0.157 57680, 585, 79077, 79132 hydrolase activity, acting on acid anhydrides, in phosphorus-containing 155066, 21, 22, 3704, 4636, 482, 38 anhydrides 738 0.012 0.157 57680, 585, 79077, 79132 39 calmodulin binding 140 0.012 0.157 hydrolase activity, acting on acid anhydrides, catalyzing transmembrane 40 movement of substances 102 0.012 0.159 155066, 21, 22, 482 41 GABA receptor activity 22 0.013 0.159 42 P-P-bond-hydrolysis-driven transmembrane transporter activity 109 0.013 0.159 155066, 21, 22, 482 43 primary active transmembrane transporter activity 109 0.013 0.159 155066, 21, 22, 482 155066, 21, 22, 3704, 4636, 482, 44 pyrophosphatase activity 735 0.013 0.16 57680, 585, 79077, 79132 45 ATPase activity, coupled to movement of substances 101 0.013 0.161 155066, 21, 22, 482 46 RNA methyltransferase activity 27 0.014 0.164 23070 47 active transmembrane transporter activity 324 0.014 0.165 155066, 21, 22, 482, 6529 48 aminoacyl-tRNA ligase activity 46 0.016 0.171 54938, 80222 49 ligase activity, forming aminoacyl-tRNA and related compounds 46 0.016 0.171 54938, 80222 50 ligase activity, forming carbon-oxygen bonds 46 0.016 0.171 54938, 80222 51 ion channel activity 366 0.017 0.178 10452 52 S-adenosylmethionine-dependent methyltransferase activity 90 0.018 0.182 23070, 6839 155066, 21, 22, 4636, 482, 57680, 53 nucleoside-triphosphatase activity 706 0.019 0.192 585, 79132 54 potassium-transporting ATPase activity 12 0.02 0.193 482

166

55 glutamate receptor activity 29 0.021 0.198 56 syntaxin-1 binding 8 0.022 0.202 57 cation:amino acid symporter activity 11 0.023 0.21 6529 58 2 iron, 2 sulfur cluster binding 17 0.024 0.214 59 cation channel activity 263 0.024 0.214 60 substrate-specific channel activity 376 0.025 0.224 10452 61 ARF GTPase activator activity 26 0.026 0.224 62 syntaxin binding 23 0.031 0.258 63 aldehyde dehydrogenase [NAD(P)+] activity 5 0.032 0.262 220 64 calcium channel activity 78 0.032 0.263 65 antibiotic transporter activity 7 0.033 0.265 66 metal ion transmembrane transporter activity 125 0.034 0.272 482, 6529 67 passive transmembrane transporter activity 393 0.036 0.273 10452 68 peptidyl-prolyl cis-trans isomerase activity 34 0.036 0.273 9360 oxidoreductase activity, acting on the aldehyde or oxo group of donors, 69 disulfide as acceptor 8 0.036 0.273 oxidoreductase activity, acting on NADH or NADPH, quinone or similar 70 compound as acceptor 44 0.036 0.273 4704 71 channel activity 392 0.039 0.285 10452 oxidoreductase activity, acting on sulfur group of donors, oxygen as 72 acceptor 6 0.04 0.287 73 hormone activity 96 0.04 0.287 2922, 5368, 7425 74 tetracycline transporter activity 6 0.041 0.292 75 E-box binding 7 0.041 0.292 7291 76 sodium ion transmembrane transporter activity 75 0.042 0.295 482, 6529 oxidoreductase activity, acting on the aldehyde or oxo group of donors, 77 NAD or NADP as acceptor 22 0.043 0.296 220, 223 78 beta-tubulin binding 19 0.043 0.299 2010, 585 79 fructose binding 6 0.048 0.324

Supplemental Table 4.5. Among genes with higher expression in Alzheimer’s, several molecular functions were enriched. P- Rank Name #Genes Value FDR SigGenes 1.33E- 11224, 2197, 25873, 6135, 6154, 6173, 6176, 1 structural constituent of ribosome 149 10 0 6188, 6193, 6207, 6222, 6232 7.58E- 2 pancreatic ribonuclease activity 10 08 0 283, 6036, 6039 2.16E- 1462, 26577, 283, 4057, 6402, 7097, 7423, 3 glycosaminoglycan binding 148 06 0.0005 79625, 929, 9388 2.64E- 4 endoribonuclease activity, producing 3'-phosphomonoesters 13 06 0.0005 283, 6036, 6039 4.03E- 5 RAGE receptor binding 7 06 0.0007 6275, 6283 9.22E- 1462, 26577, 283, 4057, 6402, 7097, 7423, 6 pattern binding 163 06 0.0014 79625, 929, 9388 1.31E- 10219, 1462, 26577, 283, 3823, 3965, 4057, 7 carbohydrate binding 339 05 0.0015 6402, 7097, 7423, 79625, 929, 9388, 9936 4.14E- 8 peptidoglycan binding 9 05 0.0043 7097, 929

endonuclease activity, active with either ribo- or deoxyribonucleic 5.51E- 9 acids and producing 3'-phosphomonoesters 18 05 0.0048 283, 6036, 6039 10 coreceptor activity 19 0.0001 0.0089 10268, 23643 11 cytokine binding 113 0.0002 0.0093 1436, 2532, 3587, 3588, 5355, 7132 12 heparin binding 110 0.0002 0.0117 26577, 283, 4057, 6402, 7423, 79625, 9388 100125288, 10049, 10346, 10488, 1050, 10848, 138151, 165, 1997, 23411, 27005, 2874, 29128, 29777, 3399, 3660, 3665, 55145, 57658, 604, 7764, 7913, 79366, 13 transcription regulator activity 923 0.0003 0.0138 84232, 9049, 9124 14 lipopolysaccharide binding 13 0.0004 0.0167 23643, 7097, 929 15 rRNA binding 24 0.0004 0.0168 283, 6135, 6222 10488, 2919, 3440, 3600, 4615, 58191, 7100, 16 cytokine receptor binding 179 0.0006 0.024 7423 17 ribonuclease activity 58 0.0022 0.0683 283, 3669, 563, 6036, 6039, 87178 18 vascular endothelial growth factor receptor binding 6 0.0022 0.0685 7423

167

1101, 11224, 1410, 2197, 24146, 25873, 4478, 4604, 6135, 6154, 6173, 6176, 6188, 6193, 6207, 6222, 6232, 81493, 81578, 19 structural molecule activity 570 0.0024 0.0707 84617, 9499 20 cyclic nucleotide-dependent protein kinase activity 9 0.0032 0.0915 5613 21 water transmembrane transporter activity 12 0.0033 0.0915 358 22 phosphate binding 13 0.0039 0.1026 10797, 4860 23 microtubule plus-end binding 10 0.0046 0.1153 81930 24 cytokine receptor activity 56 0.0051 0.1203 3587, 3588 25 metalloexopeptidase activity 37 0.0059 0.1283 10269, 10404, 165 26 death receptor activity 13 0.0066 0.1377 7132 27 monocarboxylic acid binding 57 0.007 0.1377 1592, 241, 563, 6646, 91452 28 glycosphingolipid binding 6 0.0072 0.1379 6402 29 monosaccharide binding 42 0.0076 0.1402 3965 30 proline-rich region binding 9 0.008 0.1408 4734 31 sphingolipid binding 8 0.0085 0.1464 6402 32 metallocarboxypeptidase activity 24 0.0088 0.1464 10404, 165 33 I-SMAD binding 10 0.009 0.1464 170954, 283, 4478, 4542, 4604, 54443, 34 actin binding 315 0.0094 0.1474 81930, 822, 91624, 9499 100125288, 10049, 10346, 10848, 138151, 35 transcription repressor activity 336 0.0104 0.1569 165, 1997, 23411, 2874, 3399, 604, 7764 36 ubiquitin binding 35 0.0106 0.1569 3315, 4734, 81930 37 anion binding 16 0.0109 0.1569 10797, 4860 38 IgG binding 6 0.0114 0.1569 39 voltage-gated chloride channel activity 15 0.0115 0.1569 1192, 1193 40 low-density lipoprotein receptor activity 11 0.012 0.1584 58191 41 protein complex binding 238 0.0123 0.1592 1050, 1373, 2857, 3059, 3384, 7045, 7132 42 cAMP-dependent protein kinase activity 7 0.0133 0.1602 5613 43 carboxypeptidase activity 36 0.0144 0.1662 10404, 165, 642 44 GDP binding 26 0.0152 0.171 2669, 5867 45 small conjugating protein binding 36 0.0153 0.171 3315, 4734, 81930 46 interleukin-1 receptor binding 11 0.0159 0.171 7100 47 metalloendopeptidase inhibitor activity 10 0.0165 0.171 56925, 7079 48 metalloenzyme inhibitor activity 10 0.0165 0.171 56925, 7079 49 metalloenzyme regulator activity 10 0.0165 0.171 56925, 7079 50 intramolecular oxidoreductase activity 40 0.0165 0.171 10130, 27306, 6916 51 sugar binding 173 0.0185 0.187 10219, 1462, 3823, 3965, 6402, 9936 52 endoribonuclease activity 39 0.0187 0.1871 283, 6036, 6039 53 exopeptidase activity 78 0.0205 0.1975 10269, 10404, 165, 2191, 642 54 ATPase regulator activity 7 0.0205 0.1975 10049 55 water channel activity 11 0.0208 0.1978 358 56 cyclic nucleotide-gated ion channel activity 6 0.0238 0.2144 358 57 intracellular cyclic nucleotide activated cation channel activity 6 0.0238 0.2144 358 58 scavenger receptor activity 43 0.0255 0.2238 58191, 9332 59 RNA polymerase II transcription factor activity 236 0.0261 0.2238 1050, 29128, 29777, 3660, 3665, 7913 60 regulatory region DNA binding 148 0.0263 0.2238 1050, 57658, 7764 61 transcription regulatory region DNA binding 148 0.0263 0.2238 1050, 57658, 7764 62 interferon-alpha/beta receptor binding 8 0.0306 0.2581 3440 63 collagen binding 38 0.0314 0.2613 871 64 peptidase inhibitor activity 141 0.032 0.2619 12, 25816, 56925, 6590, 7079, 710, 718, 871 65 endopeptidase inhibitor activity 133 0.0341 0.272 12, 25816, 56925, 6590, 7079, 710, 718, 871 66 tumor necrosis factor receptor activity 11 0.0355 0.2733 7132 67 alcohol transmembrane transporter activity 7 0.0365 0.2747 358 68 polyol transmembrane transporter activity 7 0.0365 0.2747 358 10346, 10488, 10848, 165, 23411, 27005, 69 transcription cofactor activity 358 0.0367 0.2747 2874, 29777, 3399, 57658, 9049, 9124 70 RNA polymerase binding 6 0.0387 0.2849 4734 71 phospholipase inhibitor activity 12 0.0389 0.2849 72 growth factor binding 111 0.0394 0.2862 3487, 3587, 3588, 4052 100125288, 10488, 1050, 1997, 23764, 73 sequence-specific DNA binding 582 0.0423 0.295 55145, 57658, 604, 7764 74 titin binding 9 0.0464 0.3166 4604 75 tumor necrosis factor binding 12 0.0466 0.3166 7132 76 cytokine activity 188 0.0487 0.3268 2919, 3440, 3600, 58191, 6398 77 lipoprotein receptor activity 15 0.0499 0.3321 58191

168

Supplemental Table 4.6. Among genes with lower expression in Alzheimer’s, the binding motifs of several transcription factors were enriched.

Rank Name #Genes P-Value FDR SigGenes 1 NRSF_01 418 0 0.022 10307, 23130, 26505, 401190, 57574, 627 2 NMYC_01 330 0.001 0.049 1846, 7291, 7425 3 CREB_01 283 0.001 0.049 148304, 1846, 9360 4 CREBP1CJUN_01 287 0.002 0.095 124044, 23070, 7425, 8364, 8366 5 MEIS1BHOXA9_02 231 0.004 0.13 1408, 627, 79680, 9360 6 AREB6_03 248 0.004 0.13 124044, 155066, 83694 7 CREB_02 137 0.005 0.13 140597, 63917, 8364, 8366 8 CREBP1_Q2 357 0.005 0.13 54984, 81576, 8364, 8366, 9360 9 NFY_01 22 0.01 0.225 10 AREB6_01 161 0.011 0.225 23130, 5368, 9379 11 ROAZ_01 326 0.024 0.384 124808, 4330, 91851 12 RFX1_02 420 0.028 0.384 124044, 145407, 1848, 29070 13 TAXCREB_02 267 0.03 0.384 115548, 79132 14 ATF_01 394 0.03 0.384 8364, 8366, 9360 15 SREBP1_01 338 0.031 0.384 140597, 26995, 57795, 627 16 MEIS1_01 101 0.042 0.484 10307 17 LYF1_01 99 0.045 0.495 23130, 242

Supplemental Table 4.7. Among genes with higher expression in Alzheimer’s, the binding motifs of several transcription factors were enriched. Rank Name #Genes P-Value FDR SigGenes 11170, 1462, 24146, 27005, 2919, 3384, 56658, 56833, 57216, 1 NFKAPPAB_01 214 0.0020 0.0951 57658, 58191, 81493, 84876 2 NKX22_01 219 0.0108 0.2252 285, 29128, 4734, 5422, 57633, 6646, 85450 10488, 1997, 25816, 4734, 55843, 56131, 58526, 6036, 712, 3 OCT_C 267 0.0146 0.2704 9531 4 PBX1_01 54 0.0300 0.3837 5166, 8334, 91947 5 HOXA3_01 183 0.0318 0.3837 2197, 4038, 648, 84173, 85450 6 CREL_01 274 0.0490 0.5000 1462, 23764, 2919, 55273, 56833, 57216, 57658, 58191, 84876

Supplemental Table 4.8. Among genes with higher expression in Alzheimer’s, the binding sites of several microRNAs were enriched. Rank Name #Genes P-Value FDR SigGenes 10269, 1050, 10848, 1462, 23593, 26511, 29128, 5355, 55227, 1 mir-506 893 0.023 0.831 57658, 58526, 604, 64089, 754, 79710 10269, 1050, 10848, 1462, 23593, 26511, 29128, 5355, 55227, 2 mir-124 890 0.035 0.831 57658, 58526, 604, 64089, 754, 79710 3 mir-433 163 0.045 0.831 256435, 284370, 91947

Supplemental Table 4.9. Among genes with lower expression in Alzheimer’s, the binding sites of several microRNAs were enriched. Rank Name #Genes P-Value FDR SigGenes 1 mir-129-5p 292 0.004 0.592 114034, 124808, 23251, 9379 2 mir-185 118 0.014 0.831 81606 3 mir-328 73 0.028 0.831 4330, 8364, 8366

169

Supplemental Table 4.10. Among genes with higher expression in Alzheimer’s, several cytogenic bands were enriched.

Rank Name #Genes P-Value FDR SigGenes 1 7p13-p12 7 8.12E-05 0.023 10268, 5224 2 17p13 30 1.85E-04 0.036 2874, 58191, 6154 3 4q25 26 2.08E-04 0.036 1950 4 10p12.31 5 3.68E-04 0.046 219681 5 1q25 21 4.45E-04 0.048 4688, 6646 6 1q23.2 7 1.09E-03 0.086 54935, 56833 7 1p36.3-p36.2 10 1.97E-03 0.122 8 1p21.3 7 2.41E-03 0.129 51375 9 17q11.2-q12 9 2.50E-03 0.129 26574 10 5q31-q33 5 3.31E-03 0.143 11 2p11.2 33 3.67E-03 0.152 822, 84173 12 22q11.1 5 0.005 0.19 128954 13 9q22 13 0.005 0.193 4814 14 Xp21.1 13 0.005 0.193 8406 15 1q21 78 0.006 0.193 6232, 6275, 6282, 6283 16 1q23 16 0.008 0.227 17 8q11 6 0.009 0.251 18 12q 12 0.012 0.301 19 9q33.2 15 0.012 0.301 92399 20 3p26.2 6 0.014 0.329 57633 21 1p31-p22 5 0.014 0.334 22802 22 7p14 7 0.024 0.464 358 23 6q24 8 0.025 0.478 23593 24 14q21 7 0.026 0.48 57161 25 5q14.3 8 0.028 0.49 1462 26 12q24.11 24 0.033 0.541 338773, 84329 27 3q24 12 0.034 0.546 28 2p16.3 7 0.038 0.604 11037 29 1p31.1 28 0.043 0.629 11080, 256435, 91624 30 6p22.1 42 0.043 0.629 29777 31 19q13.3 68 0.045 0.636 2014, 2828, 29998, 5424 32 15q22.1 6 0.049 0.661 79811

Supplemental Table 4.11. Among genes with lower expression in Alzheimer’s, several cytogenic bands were enriched. Rank Name #Genes P-Value FDR SigGenes 1 16p13.3 169 1.81E-06 0.002 10573, 124093, 1877, 21, 3049 2 10p11.2 6 7.64E-06 0.003 7587 3 8p23 6 3.45E-04 0.046 54984 4 12p12.3 25 9.67E-04 0.086 8364, 8366 5 8p21.3 21 1.03E-03 0.086 10361 6 15q25.1 11 1.81E-03 0.122 23251 7 2q21.1 15 1.86E-03 0.122 653275 8 7q31.3 11 2.61E-03 0.129 10157 9 2p22.2 6 2.70E-03 0.129 10153 10 7p14.3 8 2.83E-03 0.129 6100 11 7p11.2 8 6.16E-03 0.203 12 14q32.2 11 6.31E-03 0.203 57596 13 8p23.3 5 6.66E-03 0.206 9172 14 19q13.42 46 7.39E-03 0.221 15 10q23.32 5 0.008 0.227 16 9q34.11 42 0.011 0.286 26995 17 20q11.2-q12 6 0.016 0.354 4826 18 17p11.2 59 0.018 0.4 125170, 254272, 388341 19 2p23.1 6 0.018 0.4 81606 20 2q37.3 43 0.019 0.402 151176, 728294 21 2p25.3 10 0.019 0.402 22 14q32 16 0.021 0.416 23 5p15.33 14 0.024 0.471 65980 24 17q23 7 0.027 0.49 762 25 10q24.31 9 0.029 0.509 26 1p36.33 31 0.031 0.53 9636 27 16q24.3 34 0.032 0.536 124044, 58189 28 9q34 44 0.039 0.611 11182 29 12q23.3 17 0.042 0.629 121053 30 6p21.33 24 0.044 0.629 8364, 8366 31 19p13.3 183 0.044 0.632 5657, 566, 84717 32 14q11.2-q12 8 0.047 0.647 33 14q32.33 26 0.05 0.661

170

CHAPTER V

Conclusions and Future Directions

OVERALL SUMMARY

The goals of this dissertation are to understand (1) the regulation of homocysteine (Hcy), a one-carbon metabolite, by the heavy metal lead (Pb) and (2) the role of DNA methylation and gene expression in late onset Alzheimer’s disease (LOAD), a spontaneous neurodegenerative disease. First, in the Normative Aging Study cohort, it is demonstrated that plasma total Hcy is regulated by Pb exposure and dietary availability of vitamin B6, vitamin B12, and folate. LOAD is associated with changes in circulating Hcy, which is involved with methyl-group substrate availability for DNA methylation. Second, modest, widespread differences in DNA methylation were observed in the frontal cortex of LOAD subjects vs. controls. A follow-up study suggests that these DNA methylation marks may have functional gene expression implications. Gene expression and DNA methylation values at an individual gene were validated in additional samples. This dissertation provides the foundation for further work on the environmental influences on Hcy and epigenetics in LOAD.

171

SUMMARY OF HOMOCYSTEINE (HCY) AND LEAD (PB) EXPOSURE

Hcy is a thiol-containing intermediate in the one-carbon metabolism cycle. It is involved in folate metabolism and corresponding nucleotide synthesis, methionine metabolism and corresponding methylation of DNA, RNA, and proteins, and finally glutathione (antioxidant) synthesis (Selhub 1999). Elevations in homcysteine are associated with cardiovascular disease (CVD) (Wald et al. 2002) and neurodegeneration (Mattson and Shea 2003), including AD (Seshadri et al. 2002). This dissertation chapter shows that recent lead exposure (measured in blood) is cross-sectionally associated with Hcy levels (8.14% increase in Hcy with 3ug/dl IQR increase in blood Pb). This association was modified by vitamin B6, vitamin B12, and folate dietary factors as determined through stratified analysis. Cumulative exposure to lead (measured in tibia) was also associated with homocysteine, but this relationship disappeared after blood lead was included in the model, suggesting that blood lead mediates the association between bone lead and homocysteine.

Strengths and Weaknesses

Previous research has looked at the cross-sectional association between lead exposure and homocysteine (Chia et al. 2007; Schafer et al. 2005). Other studies have observed changes in homocysteine with diet over time (Clarke and Armitage 2000). One paper has combined lead exposure and diet in the study of homocysteine (Yakub and Iqbal 2010). This study is the first (that we are aware of) to include repeated measures of lead exposure, Hcy, and diet. It is

172

strengthened by multiple measures of lead exposure (tibia, patella, and blood) and of dietary status (plasma levels and food frequency questionnaire levels).

This study had limited power to detect interactions, due in part to missing nutrient data. In addition, results using food frequency questionnaires (FFQ) have been debated in the literature largely due to the poor correlation between FFQ and repeated dietary recall (Byers 2001; Kristal et al. 2005). In our study the consistent associations observed with concurrent measures of plasma nutrients validate the use of FFQ.

Scope and Implications of the Work

The results of this longitudinal study suggest a strong link between homocysteine and lead exposure. This dissertation did not address the downstream consequences of the association. Toxicology studies are needed to decipher the molecular mechanisms linking homocysteine and lead exposure. Potentially, Pb and Hcy may work through a common mechanism of binding and disrupting available sulfur-containing proteins (Krumdieck and Prince 2000; Needleman 2004), which may suggest synergistic or additive toxicity.

Based on the research in this dissertation, behavioral interventions to reduce blood Pb and improve dietary intake of vitamins B6 and B12 and folate will have a protective effect by lowering Hcy. Multiple randomized control trials have attempted to reduce circulating Hcy levels in people who are already ill by using dietary interventions (Mei et al. 2010). These studies have been successful in reducing circulating Hcy, but did not produce the hypothesized health benefits, specifically reductions in coronary heart disease, stroke, cardiovascular events, or all-cause mortality. Perhaps taking a public health approach with interventions prior to disease onset are needed to prevent incident cases.

173

SUMMARY OF DNA METHYLATION IN AD

The second part of the dissertation observed throughout the genome moderate changes in DNA methylation that are associated with LOAD in the frontal cortex (Bakulski et al. 2012). This study was the largest genome-wide DNA methylation investigation of the LOAD brain to date. Using 12 matched-pair LOAD case and control frontal cortex samples, we quantitatively determined DNA methylation at 27,578 CpG sites. Across 948 CpG sites that were statistically different between LOAD cases and controls after adjusting for age and sex, the mean methylation difference was 2.9%. A CpG site in the promoter for the gene Transmembrane Protein 59 (TMEM59) was 7.3% hypomethylated in LOAD cases. This gene is involved in post-translational modification of Amyloid Precursor Protein (APP), and thus β-amyloid plaque formation. We validated the DNA methylation findings using a second DNA methylation detection platform and 13 additional matched pairs. DNA methylation was associated with TMEM59 mRNA gene expression, but not with the quantity of the full-length protein. This study suggests that DNA methylation may be involved in LOAD, but future research is needed to determine the extent.

Strengths and Weaknesses

A major strength of this research was the use of well pathologically characterized human tissue from a brain region involved in LOAD. The majority of epigenetic epidemiology studies test blood DNA methylation (Foley et al. 2009). The association between circulating lymphocytes and brain epigenetics is unknown (though our research group has a funded NIEHS p30 pilot grant to investigate this question). Access to post-mortem human samples via the Michigan Alzheimer’s Disease Center provided valuable information on the in vivo inaccessible tissue.

On the other hand, a potential weakness of this research is the use of tissue samples made up of complex mixtures of cell types. Changes in composite DNA methylation may be due to changes in the percent cell-types

174

(Houseman et al. 2012). The data presented in this paper could be interpreted as actual epigenetic changes with disease, as indicators of different cellular mixtures, or as markers of past environmental exposures.

At the time, the Illumina Infinium HumanMethylation27K BeadArray was the most comprehensive and cost efficient method available. This technology queries the quantitative DNA methylation levels of 27,578 CpG sites throughout the genome with approximately two tag CpG sites per gene. Previous LOAD- DNA methylation research was based on gene-specific DNA methylation founded in a priori assumptions about the disease. The Illumina array was the most comprehensive DNA methylation array available and our study provided the most un-biased assessment of the DNA methylome of LOAD.

Genomic technology is rapidly evolving. In retrospect, with newer array and next-generation sequencing technologies available, the Illumina 27K BeadArray represents a fraction of the potential DNA methylation information that is available today. Illumina’s latest 450K array has more than fifteen times the number CpG sites and these sites have a less biased genomic distribution. The 27K array CpG sites were 92.0% in promoters and 72.5% in CpG islands. Potentially important non-genic regulation sites were largely not included in the model used in the dissertation. Studies suggest that CpG locations in gene deserts may be functionally important in the brain (Maunakea et al. 2010) and unfortunately they were unable to be captured in our study.

Scope and Implications of the Work

This research suggests DNA methylation may be involved in LOAD, but it is far from definitive. The methylation changes and sample size were modest, and results need to be confirmed in additional studies before mechanistic conclusions may be drawn. Several research groups have ongoing projects to investigate Alzheimer’s disease and epigenetics. The results of those studies combined with our research will provide a larger body of evidence on the topic. In addition, the effect sizes observed were moderate and future work will want to

175

consider separating complex cell mixtures to identify changes in neurons as a priority. Further, the findings at TMEM59 were validated through gene expression, but the full-length protein levels did not differ+ between LOAD cases and controls. The functional significance of these DNA methylation markers across the genome is presented in the next section of the dissertation.

SUMMARY OF GENE EXPRESSION AND DNA METHYLATION IN AD

The final section of the dissertation builds on the second experimental section. Using the same LOAD and neuropathologically confirmed controls as above, we observed correlations between DNA methylation differences and gene expression differences in the frontal cortex when comparing. Integrating genome-wide analyses of DNA methylation and gene expression allowed us to detect additional potential sites at low false discovery rates with LOAD associated epigenetic change. This research suggests that DNA methylation differences may have functional consequences, potentially relevant in disease.

Strengths and Weaknesses

The samples and the technologies used overlap between this dissertation chapter and the previous, therefore the strengths and weaknesses of the samples and the Illumina technology still apply. In addition, combining datasets from two companies (Illumina and Affymetrix) with multiple probesets per gene proved to be methodologically challenging. We followed a previously published pipeline to integrate the data (Sartor et al. 2011). This method proved to be successful in identifying genes for follow-up.

Scope and Implications of the Work

This research extends previous dissertation study to continue building case for DNA methylation in LOAD. Integrating gene expression and DNA methylation analyses yielded greater numbers of CpG sites for potential follow-up and mechanistic investigation. This study is an example of how DNA methylation

176

research can be enhanced when done alongside gene expression research. As studies increase the sample number of LOAD and control brains tested, this method is recommended.

FUTURE DIRECTIONS

This dissertation provides opportunities for rich potential areas of further research. Late-onset Alzheimer’s disease is an enormous public health problem and the burden will continue to increase (International 2010). LOAD is traditionally understudied in the laboratory relative to EOAD, due to the availability of EOAD analogous animal models. Despite the public health potential, within the LOAD literature, environmental risk factors and potential mechanisms of environmental toxicants are further underrepresented. Future research topics, directly related to this dissertation, include testing for a human epidemiologic link between Pb exposure and LOAD, examination of the roles of additional metals in neurodegeneration, cell-type specificity in the epigenetic molecular epidemiology, modifications to histones, and further work on dietary factors and methylation. This is a broad field of research in which could potentially contribute to our understanding and prevention of an insidious disease.

Exposure to lead has negative neurological consequences in early life (Fewtrell et al. 2004; Grosse et al. 2002). It is also associated with cognitive decline in late life (Shih et al. 2007; Weisskopf et al. 2004; Wright et al. 2003), potentially through increased hippocampal gliosis (Weisskopf et al. 2007). Though a large body of research demonstrates Pb is associated with accelerated declines in cognition, the hypothesized causal association between lead exposure and LOAD, implied in this dissertation, has not been rigorously tested. Potential epidemiologic studies to test this link are challenged by the current clinical diagnoses of AD. Clinical diagnoses of AD from 30 ADRC’s were

177

compared to post-mortem neuropathological diagnoses, with 71-87% sensitivity and 44-71% specificity (Beach et al. 2012). The ranges of clinical diagnosis of disease misclassification would require a large study size for environmental epidemiology research.

There are recent advances in neuroimaging (Matsuda and Imabayashi 2012) and circulating biomarkers (Chintamaneni and Bhaskar 2012; Doecke et al. 2012) that improve AD diagnosis and are available for research purposes. For example, the UM ADC utilizes positron emission tomography (PET) imaging methods to visualize β-amyloid in the brain and diagnose cases more accurately (Burke et al. 2011). Brain scan confirmed AD cases and controls could be measured for lead exposure at the UM Retrospective Lead Exposure Laboratory using the Cd109 K-shell X-ray Fluorescence instrument. Alternatively, post- mortem studies with neuropathologically confirmed LOAD and control samples may be able to reconstruct past metals exposure history using teeth. Researchers have identified a layer of molar tissue with relatively low turnover that dates from early development that can be quantified for metals using laser- ablation ICP-MS (Hare et al. 2011). With method validation, a case-control approach may be able to accurately reconstruct past exposure history. Finally the Pb-LOAD argument could be tested with brain post-mortem lead mapping. Demonstrating that Pb is present at the site of pathology would strengthen the argument. Post-mortem soft tissue metal content can be mapped in slices using rapid-scanning x-ray fluorescence methods (Popescu et al. 2009). Grants that propose to study the Pb-induced epigenetic changes in LOAD will need to first demonstrate the Pb-LOAD relationship in humans.

Several metals beyond Pb have been implicated in AD with varying degrees of evidence, including aluminum (Frisardi et al. 2010; Shcherbatykh and Carpenter 2007), iron (Mandel et al. 2007), copper (Brown 2009; Shcherbatykh and Carpenter 2007), zinc (Shcherbatykh and Carpenter 2007), and mercury (Gerhardsson et al. 2008). Many metals work through shared mechanisms in the

178

body. For example, lead, cadmium, mercury, and nickle bind protein sulfhydryl groups and deplete glutathione, resulting in oxidative stress (Stohs and Bagchi 1995). Future LOAD research could expand to include rigorous epidemiologic inquiry of metals in addition to Pb, using similar methods proposed above for Pb. Post-mortem exposure studies utilizing ICP-MS methods are capable of measuring up to twelve metals simultaneously. The half-life of each of these metals varies in tissues such as bone and skin, thus post-mortem exposure assessment for metals besides Pb may not reflect exposure prior to disease onset. KXRF in vivo technologies are being developed for additional metals including cadmium and arsenic (personal communication, Linda Huiling Nie, Purdue University). In vivo metals epidemiology studies coupled with early stage disease diagnosis techniques may be an effective strategy to test metals exposure as a risk factor for LOAD.

Reports have proposed pesticides (Baldi et al. 2003; Santibanez et al. 2007), solvents (Kukull et al. 1995), electromagnetic field (Sobel et al. 1996), and particulate matter in air pollution (Calderon-Garciduenas et al. 2004) may be risk factors for LOAD. The links between these exposures and LOAD have not been yet been validated. These types of environmental exposure studies have been underrepresented in the AD literature, likely due to the challenges of retrospective exposure assessment in older adults. Retrospective exposure assessment for pesticides and solvents relies heavily on questionnaire data in the absence of effective biomarkers. Future work may test the association between additional exposures and LOAD onset using long-running large cohorts where exposure information has been collected over time.

In addition to the influence of toxicants on one-carbon metabolism and epigenetics, future research may look at the direct and indirect effects of diet on the system. Results from the lead and homocysteine research paper demonstrate that dietary B6, B9, and B12 levels modify the association between blood Pb and homocysteine, an important methyl-donor for DNA methylation.

179

We have shown that DNA methylation change is associated with LOAD (Bakulski et al. 2012). Future research could test whether AD-associated DNA methylation changes are prevented with B-vitamin supplemented diet. The study of nutrient toxicant interactions is an emerging field in environmental health with potential for public health intervention.

To understand more precise mechanisms involved in LOAD pathogenesis, molecular epidemiology of epigenetics will need to refine the tissues and cell types studied. AD epigenetic epidemiology compared blood from AD cases and controls as a tissue of epidemiologic convenience (Bollati et al. 2011; Maes et al. 2007). Blood DNA methylation has not yet been tested as a biomarker of brain epigenetics. Whole brain region epigenetics is closer to the disease site, but it still represents a mixture of cell types that each likely have characteristic epigenetic profiles. Recent studies suggest that changes in circulating lymphocyte epigenetics represent shifting cell type proportions (Houseman et al. 2012). The DNA methylation differences observed in this dissertation in the human brain with LOAD may represent the understood shifting cell type proportions with disease. In future research, cell types will need to be separated from the tissue matrix. This is logistically challenging in frozen archived tissues where the cell membrane is often no longer intact. Ongoing work in the laboratory (particularly by 2nd year Ph.D. student, Zishaan Farooqui) seeks to separate nuclei of neurons from non-neurons. This allows for the study of epigenetics in neurons, and associations with environmental exposures, but it does not allow for RNA or protein functional validation. Further advancements may make the study of specific cell types more feasible.

The focus of the current research has been DNA methylation. However, recent research suggests that histones may be an important target for epigenetic modifications in the brain. Histone modifications play a role in memory formation

180

(Levenson et al. 2004) and HDAC inhibitor treatment increases memory formation (Vecsey et al. 2007). Indeed, HDAC inhibitors have been proposed for human AD clinical trials (Kazantsev and Thompson 2008) . In the male rat brain, App mRNA expression is repressed by thyroid hormone (T3) sensitive histone modifications (Belakavadi et al. 2011). T3 Treatment decreases H3K4 methylation and H3 acetylation at the App promoter. This silences App, reversed with histone deacetylase (HDAC) and histone lysine demethylase inhibitor treatment (Belakavadi et al. 2011). Histone modifications are also environmentally sensitive (Mathers et al. 2010). Together, these observations suggest that future environmental molecular epidemiology research on AD should target histone modification.

CONCLUDING REMARKS

In summary, this dissertation has provided important molecular epidemiology insights to chronic disease. It has demonstrated moderate DNA methylation differences in the LOAD brain vs. controls that may have functional gene expression consequences. In addition, homocysteine may be an important target for lead exposure toxicity and may link lead exposure to chronic disease including CVD and neurodegenerative disease. This work has implications for prevention and potential homocysteine intervention through lowering blood lead levels. This research has also spurred multiple additional research questions at the intersection of Alzheimer’s disease, one-carbon metabolism, epigenetics, and lead exposure.

181

REFERENCES

Bakulski KM, Dolinoy DC, Sartor MA, Paulson HL, Lieberman AP, Konen JR, Albin RL, Hu H, Rozek LS. 2012. Genome-wide DNA methlyation differences between late-onset Alzheimer's disease and cognitively normal controls in human frontal cortex. Journal of Alzheimer's Disease 29(2).

Baldi I, Lebailly P, Mohammed-Brahim B, Letenneur L, Dartigues JF, Brochard P. 2003. Neurodegenerative diseases and exposure to pesticides in the elderly. Am J Epidemiol 157(5): 409-414.

Beach TG, Monsell SE, Phillips LE, Kukull W. 2012. Accuracy of the clinical diagnosis of Alzheimer disease at National Institute on Aging Alzheimer Disease Centers, 2005-2010. J Neuropathol Exp Neurol 71(4): 266-273.

Belakavadi M, Dell J, Grover GJ, Fondell JD. 2011. Thyroid hormone suppression of beta-amyloid precursor protein gene expression in the brain involves multiple epigenetic regulatory events. Mol Cell Endocrinol 339(1-2): 72-80.

Bollati V, Galimberti D, Pergoli L, Dalla Valle E, Barretta F, Cortini F, Scarpini E, Bertazzi PA, Baccarelli A. 2011. DNA methylation in repetitive elements and Alzheimer disease. Brain Behav Immun.

Brown DR. 2009. Brain proteins that mind metals: a neurodegenerative perspective. Dalton Trans(21): 4069-4076.

Burke JF, Albin RL, Koeppe RA, Giordani B, Kilbourn MR, Gilman S, Frey KA. 2011. Assessment of mild dementia with amyloid and dopamine terminal positron emission tomography. Brain 134(Pt 6): 1647-1657.

Byers T. 2001. Food frequency dietary assessment: how bad is good enough? Am J Epidemiol 154(12): 1087-1088.

Calderon-Garciduenas L, Reed W, Maronpot RR, Henriquez-Roldan C, Delgado- Chavez R, Calderon-Garciduenas A, Dragustinovis I, Franco-Lira M, Aragon-Flores M, Solt AC, Altenburg M, Torres-Jardon R, Swenberg JA. 2004. Brain inflammation and Alzheimer's-like pathology in individuals exposed to severe air pollution. Toxicol Pathol 32(6): 650-658.

Chia SE, Ali SM, Lee BL, Lim GH, Jin S, Dong NV, Tu NT, Ong CN, Chia KS. 2007. Association of blood lead and homocysteine levels among lead

182

exposed subjects in Vietnam and Singapore. Occup Environ Med 64(10): 688-693.

Chintamaneni M, Bhaskar M. 2012. Biomarkers in Alzheimer's Disease: A Review. ISRN Pharmacol 2012: 984786.

Clarke R, Armitage J. 2000. Vitamin supplements and cardiovascular risk: review of the randomized trials of homocysteine-lowering vitamin supplements. Semin Thromb Hemost 26(3): 341-348.

Doecke JD, Laws SM, Faux NG, Wilson W, Burnham SC, Lam CP, Mondal A, Bedo J, Bush AI, Brown B, De Ruyck K, Ellis KA, Fowler C, Gupta VB, Head R, Macaulay SL, Pertile K, Rowe CC, Rembach A, Rodrigues M, Rumble R, Szoeke C, Taddei K, Taddei T, Trounson B, Ames D, Masters CL, Martins RN. 2012. Blood-Based Protein Biomarkers for Diagnosis of Alzheimer Disease. Arch Neurol: 1-8.

Fewtrell LJ, Pruss-Ustun A, Landrigan P, Ayuso-Mateos JL. 2004. Estimating the global burden of disease of mild mental retardation and cardiovascular diseases from environmental lead exposure. Environ Res 94(2): 120-133.

Foley DL, Craig JM, Morley R, Olsson CA, Dwyer T, Smith K, Saffery R. 2009. Prospects for epigenetic epidemiology. Am J Epidemiol 169(4): 389-400.

Frisardi V, Solfrizzi V, Capurso C, Kehoe PG, Imbimbo BP, Santamato A, Dellegrazie F, Seripa D, Pilotto A, Capurso A, Panza F. 2010. Aluminum in the diet and Alzheimer's disease: from current epidemiology to possible disease-modifying treatment. J Alzheimers Dis 20(1): 17-30.

Gerhardsson L, Lundh T, Minthon L, Londos E. 2008. Metal concentrations in plasma and cerebrospinal fluid in patients with Alzheimer's disease. Dement Geriatr Cogn Disord 25(6): 508-515.

Grosse SD, Matte TD, Schwartz J, Jackson RJ. 2002. Economic gains resulting from the reduction in children's exposure to lead in the United States. Environ Health Perspect 110(6): 563-569.

Hare D, Austin C, Doble P, Arora M. 2011. Elemental bio-imaging of trace elements in teeth using laser ablation-inductively coupled plasma-mass spectrometry. J Dent 39(5): 397-403.

Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, Wiencke JK, Kelsey KT. 2012. DNA methylation arrays as

183

surrogate measures of cell mixture distribution. BMC Bioinformatics 13(1): 86.

International AsD. 2010. World Alzheimer Report 2010: The global economic impact of dementia.

Kazantsev AG, Thompson LM. 2008. Therapeutic application of histone deacetylase inhibitors for central nervous system disorders. Nat Rev Drug Discov 7(10): 854-868.

Kristal AR, Peters U, Potter JD. 2005. Is it time to abandon the food frequency questionnaire? Cancer Epidemiol Biomarkers Prev 14(12): 2826-2828.

Krumdieck CL, Prince CW. 2000. Mechanisms of homocysteine toxicity on connective tissues: implications for the morbidity of aging. J Nutr 130(2S Suppl): 365S-368S.

Kukull WA, Larson EB, Bowen JD, McCormick WC, Teri L, Pfanschmidt ML, Thompson JD, O'Meara ES, Brenner DE, van Belle G. 1995. Solvent exposure as a risk factor for Alzheimer's disease: a case-control study. Am J Epidemiol 141(11): 1059-1071; discussion 1072-1059.

Levenson JM, O'Riordan KJ, Brown KD, Trinh MA, Molfese DL, Sweatt JD. 2004. Regulation of histone acetylation during memory formation in the hippocampus. J Biol Chem 279(39): 40545-40559.

Maes OC, Xu S, Yu B, Chertkow HM, Wang E, Schipper HM. 2007. Transcriptional profiling of Alzheimer blood mononuclear cells by microarray. Neurobiol Aging 28(12): 1795-1809.

Mandel S, Amit T, Bar-Am O, Youdim MB. 2007. Iron dysregulation in Alzheimer's disease: multimodal brain permeable iron chelating drugs, possessing neuroprotective-neurorescue and amyloid precursor protein- processing regulatory activities as therapeutic agents. Prog Neurobiol 82(6): 348-360.

Mathers JC, Strathdee G, Relton CL. 2010. Induction of epigenetic alterations by dietary and other environmental factors. Adv Genet 71: 3-39.

Matsuda H, Imabayashi E. 2012. Molecular neuroimaging in Alzheimer's disease. Neuroimaging Clin N Am 22(1): 57-65, viii.

184

Mattson MP, Shea TB. 2003. Folate and homocysteine metabolism in neural plasticity and neurodegenerative disorders. Trends Neurosci 26(3): 137- 146.

Maunakea AK, Nagarajan RP, Bilenky M, Ballinger TJ, D'Souza C, Fouse SD, Johnson BE, Hong C, Nielsen C, Zhao Y, Turecki G, Delaney A, Varhol R, Thiessen N, Shchors K, Heine VM, Rowitch DH, Xing X, Fiore C, Schillebeeckx M, Jones SJ, Haussler D, Marra MA, Hirst M, Wang T, Costello JF. 2010. Conserved role of intragenic DNA methylation in regulating alternative promoters. Nature 466(7303): 253-257.

Needleman H. 2004. Lead poisoning. Annu Rev Med 55: 209-222.

Popescu BFG, George MJ, Bergmann U, Garachtchenko AV, Kelly ME, McCrea RPE, Luning K, Devon RM, George GN, Hanson AD, Harder SM, Chapman LD, Pickering IJ, Nichol H. 2009. Mapping metals in Parkinson's and normal brain using rapid-scanning x-ray fluorescence. Physics in Medicine and Biology 54(3): 651-663.

Santibanez M, Bolumar F, Garcia AM. 2007. Occupational risk factors in Alzheimer's disease: a review assessing the quality of published epidemiological studies. Occup Environ Med 64(11): 723-732.

Schafer JH, Glass TA, Bressler J, Todd AC, Schwartz BS. 2005. Blood lead is a predictor of homocysteine levels in a population-based study of older adults. Environ Health Perspect 113(1): 31-35.

Selhub J. 1999. Homocysteine metabolism. Annu Rev Nutr 19: 217-246.

185

Shcherbatykh I, Carpenter DO. 2007. The role of metals in the etiology of Alzheimer's disease. J Alzheimers Dis 11(2): 191-205.

Sobel E, Dunn M, Davanipour Z, Qian Z, Chui HC. 1996. Elevated risk of Alzheimer's disease among workers with likely electromagnetic field exposure. Neurology 47(6): 1477-1481.

Stohs SJ, Bagchi D. 1995. Oxidative mechanisms in the toxicity of metal ions. Free Radic Biol Med 18(2): 321-336.

Vecsey CG, Hawk JD, Lattal KM, Stein JM, Fabian SA, Attner MA, Cabrera SM, McDonough CB, Brindle PK, Abel T, Wood MA. 2007. Histone deacetylase inhibitors enhance memory and synaptic plasticity via CREB:CBP-dependent transcriptional activation. J Neurosci 27(23): 6128- 6140.

Wald DS, Law M, Morris JK. 2002. Homocysteine and cardiovascular disease: evidence on causality from a meta-analysis. BMJ 325(7374): 1202.

Weisskopf MG, Hu H, Sparrow D, Lenkinski RE, Wright RO. 2007. Proton magnetic resonance spectroscopic evidence of glial effects of cumulative lead exposure in the adult human hippocampus. Environ Health Perspect 115(4): 519-523.

Weisskopf MG, Wright RO, Schwartz J, Spiro A, 3rd, Sparrow D, Aro A, Hu H. 2004. Cumulative lead exposure and prospective change in cognition among elderly men: the VA Normative Aging Study. Am J Epidemiol 160(12): 1184-1193.

Wright RO, Tsaih SW, Schwartz J, Spiro A, 3rd, McDonald K, Weiss ST, Hu H. 2003. Lead exposure biomarkers and mini-mental status exam scores in older men. Epidemiology 14(6): 713-718.

Yakub M, Iqbal MP. 2010. Association of blood lead (Pb) and plasma homocysteine: a cross sectional survey in Karachi, Pakistan. PLoS One 5(7): e11706.

186